PAGES
Subscribe Get it delivered — five to ten releases we picked each morning from the tools you run, with every release in the issue on the web. Compare plans →

The Cyber Toolchain — issue 081, September 21, 2026

THE CYBER TOOLCHAIN NO. 081
Tail
THE DAILY RELEASE FIREHOSE
PUBLISHED SEPTEMBER 21, 2026 · EVERY DAY
EDITIONS tail grep head ls diff uniq

The daily firehose — everything the toolchain shipped today, already filtered.

We read every release from the 1309 tools on our watchlist at the source — GitHub and GitLab release notes, vendor release pages and changelogs, project blogs and feeds, vendor press releases, and the source code behind the tag. Bug-fix-only releases and non-product newsroom noise are dropped; what's left is summarized down to the new capability, how to try it, and any screenshots or videos the release itself published. Every entry links to the sources it was built from.

VIEW
ISSUE VIEW scroll
INFORMATION DENSITY everything
ISSUE WIDTH reading width
Do you prefer scroll?
ADDITIONAL FILTERS 
$ tct list   # 24 tools matched
Same issue, same prompt, two writers:

What stands out across today's releases, grouped by what it lets you do. Every tool named links to its entry below.

Three unrelated things stand out. Trend Vision One's local response filters are GA, so a matching process gets killed on the Windows endpoint itself instead of waiting on a cloud round-trip. BunkerWeb adds mutual TLSmutual TLSMutual TLS, an encrypted connection where both client and server present certificates to prove their identity. It helps services verify who they are communicating with. to upstreams and a CrowdSec investigation API. NETRESEC's FlowCarp fingerprints C2 families from flow statistics, catching beacons that carry no signature.

  • respond

    Terminate a malicious process on the endpoint without a cloud round-trip

    Local response filters are now GA and auto-terminate matching processes on Windows endpoints. Containment executes on the host rather than after a detection travels to the console and a task comes back, which is the difference between stopping a loader and cleaning up after it.

    Trend Vision One

  • detect

    Catch C2 that has no signature, and hunt OT protocols in PCAP

    FlowCarp classifies C2 protocols from packet behaviour rather than byte patterns, so a family that changes its payload or moves to a fresh domain still looks the same statistically; the CLI trains custom models on your own captures. Netcap adds structured ModbusModbusA serial communication protocol originally developed by Modicon in 1979 for connecting programmable logic controllers, now widely used in industrial control systems; cyber tools target it to monitor or interact with OT environments. and S7commS7commA proprietary Siemens protocol used for communication between PLCs, HMIs, and SCADA systems in industrial control environments; cyber tools target or simulate it to audit or attack Siemens S7 device networks. parsing with CISACISAThe U.S. Cybersecurity and Infrastructure Security Agency, a federal body that publishes vulnerability advisories, threat guidance, and known-exploited-vulnerability catalogs that cyber tools use to prioritize and contextualize findings.-mapped rules, so ICSICSIndustrial Control Systems — hardware and software that monitor and control physical processes like power grids, pipelines, and manufacturing lines, making them a target for cyber tools focused on operational technology environments. traffic can be hunted as fields instead of eyeballed in Wireshark.

    NETRESEC · Netcap

  • protect

    Authenticate the reverse proxy to its own backends

    Mutual TLS to upstreams means the origin can now verify the proxy is the proxy, closing the gap where anything that reaches the backend network is trusted. The new CrowdSec investigation API and UI also let an analyst pull the decision history behind a block instead of guessing why a request was dropped.

    BunkerWeb

  • protect

    Block a bad NuGet package before it lands, and see C/C++ dependencies at all

    Package Firewall now covers NuGetNuGetMicrosoft's package manager for .NET, hosting a public registry of reusable libraries so tools can pull in dependencies without bundling or manually maintaining third-party code., so a known-malicious or policy-violating .NET package is refused at fetch time rather than found in a scan afterwards. ConanConanAn open-source C and C++ package manager that handles dependency resolution and binary caching, letting cyber tools ship reproducible builds across compilers and platforms without vendoring source. support brings C/C++ into composition analysis, a stack that traditionally has no lockfile to read and so goes uninventoried.

    Endor Labs

  • identify

    Send an externally discovered web app straight into a scanner

    EASMEASMExternal Attack Surface Management: a discipline and category of tooling that continuously discovers and monitors an organization's internet-facing assets to identify exposures before attackers do. findings can be pushed directly into DASTDASTDynamic application security testing, probing a running application from the outside for vulnerabilities. It checks behavior exposed through the application's interfaces. Essentials, removing the copy-paste step between "we found an unknown internet-facing app" and "we tested it." Attack-surface discovery only pays off when the queue it produces gets worked, and this shortens that handoff.

    Veracode

Does Opus 5 read better?
Offensive Security
◆  AI-Powered Offensive Security

Keygraph shannon

Sources Release notes → v3.3.0 Notes Rank Why this rank top features scored 28 — 28 How ranking works →

Shannon v3.3.0 adds live refresh of its AI model catalogue at scan start, so new models become available without requiring a tool update.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Shannon is an AI pentester for web applications and APIs. It analyzes your source code, identifies attack vectors, and executes real exploits to prove vulnerabilities before they reach production.

Shannon v3.3.0 adds live refresh of its AI model catalogue at scan start, so new models become available without requiring a tool update.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 1 FEATURE 0 OF 1 FEATURE most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Network refresh of AI model catalogue improved 28

Shannon now refreshes its model catalogue over the network at scan start, ensuring newly available AI models are accessible without updating the tool.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

— Describes behaviour but no mechanism, config, or command detail. v3.3.0
v3.3.0 Notes LatestStable

Shannon v3.3.0 refreshes the AI model catalogue over the network at scan start so new models are available without a tool update.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch v3.3.0 https://github.com/KeygraphHQ/shannon.git
# already have the repo? check out this version:
$ git checkout v3.3.0
Was this useful?
Application Security
◆  SAST / SCA / Code Security

cve-lite-cli

Sources Release notes → v1.36.0 Notes Rank Why this rank top features scored 50/10 — 30 How ranking works →

cve-lite-cli v1.36.0 opens up its PR delta renderer for external use and patches two false-result bugs from the prior release.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Fast, developer-friendly JS/TS dependency vulnerability scanner with local lockfile scanning, OSV matching, direct vs transitive visibility, --fix, JSON output, and practical remediation guidance.

cve-lite-cli v1.36.0 opens up its PR delta renderer for external use and patches two false-result bugs from the prior release.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 2 FEATURES 0 OF 2 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Public PR delta renderer package new 50

The PR delta renderer, which generates pull request comments by diffing two scan reports, is now published as part of the distributed cve-lite-cli package.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ This lets external tools and CI workflows outside the repository generate PR comments from scan diffs without reimplementing the logic.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

— Names the module and its function but no CLI flag or command v1.36.0
02 00
False-result bug fixes improved 10

Fixes two false-result bugs that were introduced in v1.35.0; the specific bugs are not described.

— No details on which false results or how fixed v1.36.0
v1.36.0 Notes LatestStable

cve-lite-cli v1.36.0 ships the PR delta renderer as a public package and fixes two false-result bugs from v1.35.0.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch v1.36.0 https://github.com/OWASP/cve-lite-cli.git
# already have the repo? check out this version:
$ git checkout v1.36.0
  • Publishes the PR delta renderer as part of the distributed package, enabling external tools and CI workflows outside this repository to generate pull request comments from two scan reports.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

1 detail in Everything

Was this useful?

Aikido Security

Sources Blog post → announced · Sep 21, 2026 Blog Rank Why this rank top features scored 83, +10 product media — 93 How ranking works →

Aikido released Altar, an open-weight AI security model built to run Aikido Machine's autonomous pentesting entirely inside customer-controlled or air-gapped infrastructure.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Aikido Security is a developer-focused security platform that identifies and remediates vulnerabilities in code and dependencies.

Aikido released Altar, an open-weight AI security model built to run Aikido Machine's autonomous pentesting entirely inside customer-controlled or air-gapped infrastructure.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 1 FEATURE 1 OF 1 FEATURE most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 01
Altar open-weight AI security model new 83

Altar is an open-weight security AI model (328 GB, W4A16 quantizationquantizationStoring a model's weights at lower numeric precision, such as 8 or 4 bits instead of 16, so it needs less memory and runs faster. It trades a small amount of accuracy for fitting on smaller hardware., 168 of 256 routed experts retained) that powers Aikido Machine's autonomous pentesting entirely inside customer-controlled or air-gapped infrastructure, without sending code or findings to a third-party inference service.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ It is deployable on a 4-H200s node using vLLM, with model weightsmodel weightsThe numerical values a model learns during training. Loading them into the model's architecture restores its learned behavior for inference or further training., a model card, license, serving flags, and deployment instructions published under Aikido's organization.

— Names exact specs and deployment path, but no runnable command shown. launch-20260921-b3ced206
announced · Sep 21, 2026 Blog

Aikido releases Altar, an open-weight AI security model for air-gapped and sovereign deployments inside Aikido Machine.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Introduces Altar, an open-weight security AI model (328 GB, W4A16 quantizationquantizationStoring a model's weights at lower numeric precision, such as 8 or 4 bits instead of 16, so it needs less memory and runs faster. It trades a small amount of accuracy for fitting on smaller hardware., 168 of 256 routed experts retained) that powers Aikido Machine's autonomous pentesting entirely inside customer-controlled or air-gapped infrastructure without sending code or findings to a third-party inference service.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Altar is deployable on a 4-H200s node using vLLM, with model weightsmodel weightsThe numerical values a model learns during training. Loading them into the model's architecture restores its learned behavior for inference or further training., a model card, license, serving flags, and deployment instructions available under Aikido's organization.

2 details in Everything

Was this useful?

Veracode

Sources Blog / feed → EASM updates (2026-09-21) Blog Rank Why this rank top features scored 75/50/45, +6 runnable example — 62.67 How ranking works →

Veracode's EASMEASMExternal Attack Surface Management: a discipline and category of tooling that continuously discovers and monitors an organization's internet-facing assets to identify exposures before attackers do. module gets a full UI redesign, a new Findings view, Dashboard widgets, Business Units management, and a direct path to push discovered web apps into DASTDASTDynamic application security testing, probing a running application from the outside for vulnerabilities. It checks behavior exposed through the application's interfaces. Essentials for scanning, alongside real-time scan notifications, multi-project visibility, scan scheduling and AI-driven risk prioritization.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Veracode is a cloud-based application security platform that performs static, dynamic, and software composition analysis to identify vulnerabilities in code.

Veracode's EASMEASMExternal Attack Surface Management: a discipline and category of tooling that continuously discovers and monitors an organization's internet-facing assets to identify exposures before attackers do. module gets a full UI redesign, a new Findings view, Dashboard widgets, Business Units management, and a direct path to push discovered web apps into DASTDASTDynamic application security testing, probing a running application from the outside for vulnerabilities. It checks behavior exposed through the application's interfaces. Essentials for scanning, alongside real-time scan notifications, multi-project visibility, scan scheduling and AI-driven risk prioritization.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 6 FEATURES 0 OF 6 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Export EASM assets to DAST Essentials as scan targets new 75

Enables export of web applications from EASM directly into DAST Essentials by marking assets as DAST Candidates, which creates discovered targets in DAST Essentials ready for authenticated scanning.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Promote a discovered web application to a DAST scan target without leaving EASM — useful when an asset surfaces during footprint discovery and you want to immediately queue it for authenticated scanning.
📍In the EASM UI, open the asset you want to scan, mark it as a DAST Candidate, then navigate to DAST Essentials to find the asset listed as a discovered target ready for scanning.
— Named mechanism and step-by-step usage example provided EASM updates (2026-09-21)
thinner coverage below
02 00
Dashboard widgets: Heatmap, World map, Things new 50

Adds Dashboard widgets — Heatmap, World map, and Things — with filterable data views and a shortcut to create reports.

— Names the three widgets but not the filter options EASM updates (2026-09-21)
03 00
Real-time notifications, visibility, scheduling and AI prioritization in EASM new 45

Adds real-time in-app notifications when scans complete, fail, or are cancelled; multi-project visibility for centralized risk management across all projects; scan scheduling to streamline orchestration; AI-powered intelligence for accelerated risk prioritization; and an Issues capability to streamline remediation efforts and improve focus on cyber hygiene.

— Five thin one-line additions grouped, none with a named UI path or config EASM updates (2026-09-21)
04 00
Findings view in EASM new 45

Adds a Findings view to surface large volumes of security data as actionable insights, with prioritization workflows and expanded reporting.

— Describes purpose but not exact fields or filters EASM updates (2026-09-21)
05 00
Redesigned EASM UI with Legacy UI option improved 40

Introduces a redesigned EASM UI with improved navigation and usability. The Legacy UI remains available for users who aren't ready to switch.

— Names the change but no navigation path or scope detail EASM updates (2026-09-21)
06 00
Business Units management in EASM new 35

Adds Business Units management, enabling creation, assignment, and domain association.

— States capability without mechanism or UI location EASM updates (2026-09-21)
EASM updates (2026-09-21) Blog LatestStable

Veracode EASMEASMExternal Attack Surface Management: a discipline and category of tooling that continuously discovers and monitors an organization's internet-facing assets to identify exposures before attackers do. gains redesigned UI, real-time scan notifications, multi-project visibility, scan scheduling, and AI-powered intelligence.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Promote a discovered web application to a DASTDASTDynamic application security testing, probing a running application from the outside for vulnerabilities. It checks behavior exposed through the application's interfaces. scan target without leaving EASM — useful when an asset surfaces during footprint discovery and you want to immediately queue it for authenticated scanning.
📍In the EASM UI, open the asset you want to scan, mark it as a DAST Candidate, then navigate to DAST Essentials to find the asset listed as a discovered target ready for scanning.
  • Adds real-time in-app notifications when scans complete, fail, or are cancelled.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds multi-project visibility for centralized risk management across all projects.
  • Adds scan scheduling to streamline scan orchestration across projects.
  • Adds AI-powered intelligence for accelerated risk prioritization.
  • Introduces redesigned EASM UI with improved navigation and usability, with the Legacy UI still available.
+5 more show less
  • Adds a Findings view to surface large volumes of security data as actionable insights with prioritization workflows and expanded reporting.
  • Adds Issues capability to streamline remediation efforts and improve focus on cyber hygiene.
  • Adds Dashboard widgets — Heatmap, World map, and Things — with filterable data views and a shortcut to create reports.
  • Adds Business Units management, enabling creation, assignment, and domain association.
  • Enables export of web applications from EASM directly into DAST Essentials by marking assets as DAST Candidates to create discovered targets for DAST scanning.
  • Adds real-time in-app notifications when scans complete, fail, or are cancelled.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds multi-project visibility for centralized risk management across all projects.
  • Adds scan scheduling to streamline scan orchestration across projects.
  • Adds AI-powered intelligence for accelerated risk prioritization.
  • Adds a Findings view to surface large volumes of security data as actionable insights with prioritization workflows and expanded reporting.
+3 more show less
  • Adds Issues capability to streamline remediation efforts and improve focus on cyber hygiene.
  • Adds Business Units management, enabling creation, assignment, and domain association.
  • Enables export of web applications from EASM directly into DAST Essentials by marking assets as DAST Candidates to create discovered targets for DAST scanning.

2 more in Everything

Was this useful?
◆  Secrets Detection & Management

infisical

Sources Release notes → v0.165.14 Notes Rank Why this rank top features scored 52/48/30 — 43.33 How ranking works →

Infisical's latest release adds PowerDNS as a DNS-01DNS-01An ACME protocol challenge type that proves domain ownership by having the requester publish a specific TXT record in DNS, letting cyber tools obtain TLS certificates for hosts that may not be publicly reachable over HTTP. provider for automated certificate issuance, introduces fine-grained request routing controls for Agent Vault, and adds event-triggered alerts alongside billing and secrets-UI improvements.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Infisical is the open-source platform for secrets, certificates, and privileged access management.

Infisical's latest release adds PowerDNS as a DNS-01DNS-01An ACME protocol challenge type that proves domain ownership by having the requester publish a specific TXT record in DNS, letting cyber tools obtain TLS certificates for hosts that may not be publicly reachable over HTTP. provider for automated certificate issuance, introduces fine-grained request routing controls for Agent Vault, and adds event-triggered alerts alongside billing and secrets-UI improvements.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 5 FEATURES 0 OF 5 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Fine-grained routing rules for Agent Vault new 52

Adds method and path rules, custom headers, and substitutions to Agent Vault for fine-grained request routing and transformation control.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

— Lists concrete rule types but no exact syntax or flags. v0.165.14
02 00
PowerDNS DNS-01 provider for ACME new 48

Adds PowerDNS as a DNS-01 provider for ACMEACMEAn IETF-standard protocol (RFC 8555) for automating TLS certificate issuance and renewal, letting cyber tools manage certificates without manual CA interaction. certificate authorities, enabling automated certificate issuance via PowerDNS-managed zones.

— Names provider and mechanism but no config steps. v0.165.14
03 00
Event outbox and event-triggered alerts new 30

Adds a transactional event outbox and event-triggered alerts for automated notifications on secret or infrastructure events.

— Describes capability but no event types or config named. v0.165.14
04 00
Bulk secret selection and clipboard copy improved 26

Simplifies bulk secret selection and adds clipboard copying to the secrets UI.

— UI change described but no navigation path given. v0.165.14
05 00
Billing v2 usage breakdown and org listing new 23

Adds billing v2 usage breakdown and organization listing via a new API implementation.

— No endpoint, fields, or usage details given. v0.165.14
v0.165.14 Notes Stable

Infisical v0.165.14 adds PowerDNS DNS-01DNS-01An ACME protocol challenge type that proves domain ownership by having the requester publish a specific TXT record in DNS, letting cyber tools obtain TLS certificates for hosts that may not be publicly reachable over HTTP. support, Agent Vault routing rules, and event-triggered alerts⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch v0.165.14 https://github.com/Infisical/infisical.git
# already have the repo? check out this version:
$ git checkout v0.165.14
  • Adds PowerDNS as a DNS-01 provider for ACMEACMEAn IETF-standard protocol (RFC 8555) for automating TLS certificate issuance and renewal, letting cyber tools manage certificates without manual CA interaction. certificate authorities, enabling automated certificate issuance via PowerDNS-managed zones.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds method and path rules, custom headers, and substitutions to Agent Vault for fine-grained request routing and transformation control.
  • Adds a transactional event outbox and event-triggered alerts for automated notifications on secret or infrastructure events.
  • Adds billing v2 usage breakdown and organization listing via a new API implementation.
  • Simplifies bulk secret selection and adds clipboard copying to the secrets UI.
  • Adds PowerDNS as a DNS-01 provider for ACMEACMEAn IETF-standard protocol (RFC 8555) for automating TLS certificate issuance and renewal, letting cyber tools manage certificates without manual CA interaction. certificate authorities, enabling automated certificate issuance via PowerDNS-managed zones.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds a transactional event outbox and event-triggered alerts for automated notifications on secret or infrastructure events.
  • Adds billing v2 usage breakdown and organization listing via a new API implementation.

2 more in Everything

Was this useful?
◆  Web Application Security

Bunkerity BunkerWeb

Sources Release notes → v1.6.15 Notes Rank Why this rank top features scored 95/95/75, +15 breaking change, +10 product media, +6 runnable example — 119.33 How ranking works →

Open-source and cloud-native Web Application Firewall (WAF)

BunkerWeb 1.6.15 ships a guided, flag-rich Docker upgrade script with automatic backup, health checks and rollback; a full CrowdSec investigation API and UI; mutual TLSmutual TLSMutual TLS, an encrypted connection where both client and server present certificates to prove their identity. It helps services verify who they are communicating with. to upstreams; and a bundle of new file/IP-reputation plugins (SentinelOne, SysWarden, ClamAV, VirusTotal), alongside Redis-resilient session storage and per-setting draft mode in the Web UI RAW editor.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 12 FEATURES 1 OF 12 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Guided Docker upgrade script for BunkerWeb stacks new 95

install-bunkerweb.sh --docker gains --compose-dir PATH (stack directory instead of cwd), -v, --version VERSION and --image-tag TAG to select the target image, --backup-dir PATH for the pre-upgrade backup archive, --overwrite-compose to force-regenerate docker-compose.yml, --force-type-change for destructive topology switches, --no-pull to skip image pulls, and --docker-wait-timeout N (default 600s) for health-check patience.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Before changes it runs bwcli plugin backup save in the Scheduler container and backs up .env and docker-compose.yml as a .bak.<timestamp> copy, aborting on backup failure unless overridden; it reads the running version from the live container to reliably handle rolling tags (latest, testing) and interrupted upgrades; it rewrites .env preserving manually added entries and only regenerates docker-compose.yml if it still matches the script-generated original; it auto-detects install type, secrets, host ports, worker list and Compose project name from .env to prevent accidental credential rotation or port changes; it then runs docker compose pull and docker compose up -d, recreating only containers whose image changed, requires two healthy observations at least ten seconds apart, exits with status 2 (no rollback) on timeout, backs up the database, and refuses downgrades.

Upgrade while preserving local docker-compose.yml edits but skip pulling images (useful when images are pre-pulled in an air-gapped environment).
$ bash install-bunkerweb.sh --docker --compose-dir /opt/bunkerweb -v 1.6.15 --no-pull
Force-regenerate docker-compose.yml when switching stack topology (e.g. from all-in-one to split manager/worker), accepting the destructive topology change.
$ bash install-bunkerweb.sh --docker --compose-dir /opt/bunkerweb -v 1.6.15 --overwrite-compose --force-type-change
Upgrade a stack that lives outside the current directory to a specific version, with a custom backup destination and an extended health-check timeout for slow first starts.
$ bash install-bunkerweb.sh --docker --compose-dir /opt/bunkerweb --version 1.6.15 --backup-dir /mnt/backups --docker-wait-timeout 900
Force regeneration of a locally modified docker-compose.yml when the local edits are no longer needed and you want the script-generated layout back.
$ bash install-bunkerweb.sh --docker --compose-dir /opt/bunkerweb --version 1.6.15 --overwrite-compose
— Every flag named with runnable upgrade commands and rollback mechanics. v1.6.15
02 00
CrowdSec investigation and management API/UI new 95

Adds CROWDSEC_MANAGEMENT_LOGIN and CROWDSEC_MANAGEMENT_PASSWORD config keys for a dedicated CrowdSec machine identity, plus endpoints GET /crowdsec (connection status/sync/errors), GET /crowdsec/{connection_id}/decisions (filterable by origin and scenario, paginated via offset/limit, max 200), GET /crowdsec/{connection_id}/ips/{ip} (up to 200 decisions, 50 alerts, 50 reports with totals), GET /crowdsec/{connection_id}/alerts/{alert_id}, GET /crowdsec/{connection_id}/allowlists (paginated, up to 200 entries per list), GET /crowdsec/{connection_id}/allowlists/check?ip={ip}, and DELETE /crowdsec/{connection_id}/decisions/{decision_id} (requires scope, value, decision_type in the JSON body). New delegated permissions crowdsec_read and crowdsec_delete scope this under the bans resource, and a new Additional Pages → CrowdSec UI page consolidates active decisions, alerts, BunkerWeb reports, local bans and native allowlist status with pre-filled IP lookup links.

Investigate all active CrowdSec decisions for a specific IP across a connection, to confirm whether a block is still in force before escalating a support ticket.
$ curl -s -H 'Authorization: Bearer <api_token>' 'https://bunkerweb.example.com/api/crowdsec/<connection_id>/ips/203.0.113.42'
Remove a mistaken CrowdSec ban on an IP so it propagates to all bouncers at their next stream update.
$ curl -s -X DELETE -H 'Authorization: Bearer <api_token>' -H 'Content-Type: application/json' -d '{"scope":"ip","value":"203.0.113.42","decision_type":"ban"}' 'https://bunkerweb.example.com/api/crowdsec/<connection_id>/decisions/<decision_id>'
Remove an active CrowdSec ban from the local API without touching BunkerWeb-local bans — useful when a decision was issued in error on a shared engine.
$ curl -X DELETE https://bunkerweb.example.com/crowdsec/<connection_id>/decisions/<decision_id> \
  -H 'Authorization: Bearer <api_token>' \
  -H 'Content-Type: application/json' \
  -d '{"scope": "ip", "value": "203.0.113.42", "decision_type": "ban"}'
Investigate all active CrowdSec decisions for a suspicious IP, including matching alerts and retained BunkerWeb reports, to triage an incident.
$ curl -s 'https://bunkerweb.example.com/crowdsec/<connection_id>/ips/203.0.113.42' \
  -H 'Authorization: Bearer <api_token>' | jq .
Programmatically remove a specific CrowdSec ban on an IP address via the public API, useful in an automated remediation workflow.
$ curl -s -X DELETE https://bunkerweb.example.com/api/crowdsec/<connection_id>/decisions/<decision_id> \
  -H 'Authorization: Bearer <api_token>' \
  -H 'Content-Type: application/json' \
  -d '{"scope": "ip", "value": "198.51.100.42", "decision_type": "ban"}'
Investigate a suspicious IP across all CrowdSec decisions, alerts, and retained BunkerWeb reports in one call.
$ curl -s 'https://bunkerweb.example.com/api/crowdsec/<connection_id>/ips/198.51.100.42' \
  -H 'Authorization: Bearer <api_token>'
— Every endpoint, method and required field named with runnable curl calls. v1.6.15
03 00
Session storage, Redis fallback and revocation controls new 75

UI_USE_REDIS=no decouples the Web UI from Redis independently of the global USE_REDIS setting; when Redis is unavailable or rejects writes due to maxmemory, UI sessions fall back to a local per-host cache, with logout and session-ID rotation revocations recorded in both stores. Destroyed sessions are rejected on next use, locally or cluster-wide when Redis is enabled, and the new SESSIONS_REVOCATION_MEMORY_SIZE config key controls the in-memory store size for revoked cookie session identifiers.

Disable Redis for the BunkerWeb UI session store only, keeping Redis active for inter-instance lock and report sharing.
ini 
UI_USE_REDIS=no
Isolate the UI from Redis session storage without disabling Redis-backed bannishments and inter-instance reporting — useful when you want Redis for data sharing but local session handling.
ini 
UI_USE_REDIS=no
Isolate the UI's Redis dependency from the rest of BunkerWeb — useful when you want bans and reports shared via Redis but need the UI to run without it.
env 
UI_USE_REDIS=no
— Named config keys and fallback mechanism, no numeric size default given. v1.6.15
04 00
Header-based bypass and match rules across security plugins new 75

Adds numbered header-match config pairs ANTIBOT_IGNORE_HEADER_NAME/ANTIBOT_IGNORE_HEADER_VALUE, BLACKLIST_HEADER_NAME/BLACKLIST_HEADER_VALUE, BLACKLIST_IGNORE_HEADER_NAME/BLACKLIST_IGNORE_HEADER_VALUE, and COUNTRY_IGNORE_HEADER_NAME/COUNTRY_IGNORE_HEADER_VALUE (with PCRE-matched values) to bypass or trigger the antibot, blacklist and country plugins; header-based matching is also extended to the whitelist, greylist and dnsbl plugins, with header values stored as password settings that are never logged or cached.

Allow an internal health-check service to bypass the antibot challenge by presenting a shared secret header.
env 
ANTIBOT_IGNORE_HEADER_NAME_1=X-Internal-Token
ANTIBOT_IGNORE_HEADER_VALUE_1=^my-secret-token$
Allow internal monitoring probes to skip the antibot challenge by presenting a shared secret header, without opening access to all traffic.
ini 
ANTIBOT_IGNORE_HEADER_NAME_1=X-Internal-Token
ANTIBOT_IGNORE_HEADER_VALUE_1=^my-secret-rotation-value$
Allow a trusted internal service to bypass the antibot challenge by presenting a shared-secret header, while anchoring the regex so partial matches are rejected.
env 
ANTIBOT_IGNORE_HEADER_NAME_1=X-Internal-Token
ANTIBOT_IGNORE_HEADER_VALUE_1=^my-secret-token$
— All header config keys enumerated with a working config example. v1.6.15
05 00
Per-setting draft mode in RAW editor new 65

In the RAW editor, prefix a KEY=value line with # to mark it draft (value stored but not applied, shown with a muted # prefix, inherited/default value takes effect); press Backspace on a draft line to reactivate it. Draft lines persist across saves in RAW mode and are preserved when saving in Simple or Advanced mode.

Stage a config change in the RAW editor without applying it yet — useful for preparing a value to activate in a single future save.
📍# In the BunkerWeb web UI RAW editor: # 1. Open a service or global settings RAW editor. # 2. Place the cursor at the start of a KEY=value line. # 3. Press # to toggle the line into draft state (line becomes highlighted). # 4. Save — the value is stored but the effective config remains unchanged. # 5. Press Backspace on the draft line (or # again) to activate, then save to apply.
Prepare a config change in the RAW editor without applying it yet — stage it as a draft and activate it in a later save when ready.
📍In the RAW editor, place the cursor at the start of a KEY=value line and press # to toggle it to draft state. The line is highlighted and the value is stored but not applied. Press Backspace on the drafted line to activate it, then save to apply.
Temporarily disable a setting in RAW mode without deleting it, so its inherited or default value applies while you keep the line for later activation.
📍In the RAW mode editor, navigate to the start of a setting line and press # to mark it as draft. The line gains a muted # prefix and the inherited/default value takes effect. Press Backspace at the start of the line to reactivate it, then click Save to persist.
— Exact keystrokes given as UI steps, no config-file equivalent. v1.6.15
06 00
New plugin bundle: SentinelOne, SysWarden, ClamAV, VirusTotal new 60

EXTERNAL_PLUGIN_URLS installs the v1.12 plugin bundle from a URL such as https://github.com/bunkerity/bunkerweb-plugins/archive/refs/tags/v1.12.zip. bunkerweb-plugins/sentinelone checks uploaded file hashes and client IPs against the SentinelOne API, blocking on reputation or IOCsIOCsAn indicator of compromise, an observable clue that may point to malicious activity, such as a suspicious file hash or domain. It is evidence to investigate, rather than proof by itself.; bunkerweb-plugins/syswarden syncs BunkerWeb bans to SysWarden for kernel-level nftables blocking and imports its block/allow lists; a ClamAV plugin scans uploaded files and rejects requests when malicious; a VirusTotal plugin checks uploaded file hashes and client IPs against the VirusTotal API, blocking when configured reputation thresholds are exceeded.

Install the full v1.12 plugin bundle — including SentinelOne, SysWarden, ClamAV, and VirusTotal — by pointing the scheduler at the tagged archive.
text 
EXTERNAL_PLUGIN_URLS=https://github.com/bunkerity/bunkerweb-plugins/archive/refs/tags/v1.12.zip
— Plugins and install key named, each mechanism only one line. product docs
07 00
ModSecurity audit log path and concurrent logging new 60

Adds MODSECURITY_SEC_AUDIT_LOG config key to configure the ModSecurityModSecurityAn open-source web application firewall engine, maintained by the OWASP community, that inspects HTTP traffic against rule sets to block malicious requests before they reach an application. audit log path (under /var/log/bunkerweb by default), and adds opt-in concurrent audit logging for ModSecurity with instance-side storage validation.

Send ModSecurity concurrent audit log entries to a dedicated path for ingestion by a SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents., rather than the default log location.
ini 
MODSECURITY_SEC_AUDIT_LOG=/var/log/bunkerweb/modsec_audit.log
— Named config key with a concrete example path, but validation detail is thin. v1.6.15
thinner coverage below
08 00
Mutual TLS to upstreams in reverseproxy and grpc plugins new 50

Supports presenting a client certificate to upstreams for mutual TLS in the reverseproxy and grpc plugins, with configurable upstream CRLCRLA Certificate Revocation List is a signed file, published by a certificate authority, listing certificates it has revoked before expiry so that clients can reject compromised or mis-issued credentials without contacting the CA live., TLS protocols, and ciphers validated and distributed by the Scheduler.

— Mechanism named but no specific config key given. v1.6.15
09 00
gRPC plugin feature parity with reverseproxy improved 50

Expands the grpc plugin to match reverseproxy configurability, including client response headers, pass/ignore headers, underscores in headers, buffer and body-size controls, auth_request, and upstream TLS verification.

— Feature list named but no exact config keys given. v1.6.15
10 00
Per-URL request body size limit new 50

Adds per-URL REVERSE_PROXY_MAX_CLIENT_SIZE config key to control both the request body size and the ModSecurity body-size limit on a per-location basis.

— Named config key but no example or default value given. v1.6.15
11 01
Two-factor auth anti-replay moved to shared database breaking 50

Replay protection for two-factor authentication codes is now stored in the shared database across Web UI replicas instead of per-replica memory. During a rolling upgrade, a TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip. code already displayed may be rejected for up to 33 seconds; upgrade all Web UI replicas simultaneously to minimise the window, and TOTP encryption keys must remain available to every replica as before.

— Migration guidance given but no config key or command. product docs
12 00
logrotate integration for all-in-one image new 45

Adds logrotate integration for the all-in-one image; mount /etc/logrotate.d/bunkerweb to provide a custom rotation policy.

— Mount path given but no example policy or defaults. v1.6.15
└──▷ ALSO FROM THESE RELEASES
Disable Redis for the Web UI only — keeping global ban and report sharing intact — when the UI workers are hitting Redis connection limits.
ini 
UI_USE_REDIS=no
└──▷ BREAKING ON UPGRADE
  • !
    Two-factor authentication anti-replay state is now stored in the shared database instead of per-replica memory; during a rolling upgrade, a TOTP code already displayed may be rejected for up to 33 seconds — upgrade all web UI replicas simultaneously to minimise the window. TOTP encryption keys must remain available to every replica as before.
  • !
    Two-factor authentication replay protection is now stored in the shared database and only covers replicas running this version; during a mixed-version migration window an already-displayed TOTP code may be rejected for up to 33 seconds.
v1.6.15 Notes LatestStable

BunkerWeb 1.6.15 adds mTLSmTLSMutual TLS, an encrypted connection where both client and server present certificates to prove their identity. It helps services verify who they are communicating with. to upstreams, per-header allow/block criteria, per-URL client size limits, and a syslog-ngsyslog-ngAn open-source log collector and router that ingests, filters, parses, and forwards log messages from many sources to many destinations, giving cyber tools a flexible, high-throughput pipeline for centralizing event data. UI log collector.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch v1.6.15 https://github.com/bunkerity/bunkerweb.git
# already have the repo? check out this version:
$ git checkout v1.6.15
└──▷ USE IT
Disable Redis for the Web UI only — keeping global ban and report sharing intact — when the UI workers are hitting Redis connection limits.
/etc/bunkerweb/variables.env 
UI_USE_REDIS=no
Send ModSecurityModSecurityAn open-source web application firewall engine, maintained by the OWASP community, that inspects HTTP traffic against rule sets to block malicious requests before they reach an application. concurrent audit log entries to a dedicated path for ingestion by a SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents., rather than the default log location.
/etc/bunkerweb/variables.env 
MODSECURITY_SEC_AUDIT_LOG=/var/log/bunkerweb/modsec_audit.log
  • Adds UI_USE_REDIS=no to disable Redis exclusively for the Web UI without disabling the global Redis-backed ban and report sharing.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds MODSECURITY_SEC_AUDIT_LOG to configure the ModSecurity audit log path under /var/log/bunkerweb.
  • Adds per-URL REVERSE_PROXY_MAX_CLIENT_SIZE to control both request body size and the ModSecurity body-size limit on a per-location basis.
  • Supports presenting a client certificate to upstreams for mutual TLSmutual TLSMutual TLS, an encrypted connection where both client and server present certificates to prove their identity. It helps services verify who they are communicating with. in the reverseproxy and grpc plugins, with configurable upstream CRLCRLA Certificate Revocation List is a signed file, published by a certificate authority, listing certificates it has revoked before expiry so that clients can reject compromised or mis-issued credentials without contacting the CA live., TLS protocols, and ciphers validated and distributed by the Scheduler.
  • Adds header-based matching to antibot, blacklist, whitelist, greylist, dnsbl, and country plugins — match on a request header name and optional PCRE value regex as an ignore or list criterion; header values are stored as password settings and are never logged or cached.
+7 more show less
  • Expands the grpc plugin to match reverseproxy configurability, including client response headers, pass/ignore headers, underscores in headers, buffer and body-size controls, auth_request, and upstream TLS verification.
  • Adds CrowdSec IP investigation, allowlist inspection, and decision removal through the Web UI and API, with richer report details and on-page allowlist guidance.
  • Adds opt-in concurrent audit logging for ModSecurity with instance-side storage validation.
  • Adds install-bunkerweb.sh --docker upgrade capability for Docker stacks previously generated by the installer, backing up the database and preserving secrets, ports, and Compose edits while refusing downgrades.
  • Adds logrotate integration for the all-in-one image; mount /etc/logrotate.d/bunkerweb to provide a custom rotation policy.
  • Adds destroyed-session rejection on next use, locally or cluster-wide when Redis is enabled.
  • Allows individual settings to remain as drafts in the RAW editor, with explicit activation and inherited/default effective values.
  • Adds UI_USE_REDIS=no to disable Redis exclusively for the Web UI without disabling the global Redis-backed ban and report sharing.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds MODSECURITY_SEC_AUDIT_LOG to configure the ModSecurity audit log path under /var/log/bunkerweb.
  • Adds per-URL REVERSE_PROXY_MAX_CLIENT_SIZE to control both request body size and the ModSecurity body-size limit on a per-location basis.
  • Supports presenting a client certificate to upstreams for mutual TLSmutual TLSMutual TLS, an encrypted connection where both client and server present certificates to prove their identity. It helps services verify who they are communicating with. in the reverseproxy and grpc plugins, with configurable upstream CRLCRLA Certificate Revocation List is a signed file, published by a certificate authority, listing certificates it has revoked before expiry so that clients can reject compromised or mis-issued credentials without contacting the CA live., TLS protocols, and ciphers validated and distributed by the Scheduler.
  • Adds header-based matching to antibot, blacklist, whitelist, greylist, dnsbl, and country plugins — match on a request header name and optional PCRE value regex as an ignore or list criterion; header values are stored as password settings and are never logged or cached.
+5 more show less
  • Adds opt-in concurrent audit logging for ModSecurity with instance-side storage validation.
  • Adds install-bunkerweb.sh --docker upgrade capability for Docker stacks previously generated by the installer, backing up the database and preserving secrets, ports, and Compose edits while refusing downgrades.
  • Adds logrotate integration for the all-in-one image; mount /etc/logrotate.d/bunkerweb to provide a custom rotation policy.
  • Adds destroyed-session rejection on next use, locally or cluster-wide when Redis is enabled.
  • Allows individual settings to remain as drafts in the RAW editor, with explicit activation and inherited/default effective values.

2 more in Everything

Was this useful?

PortSwigger Burp Suite

Sources Blog / feed → Professional / Community 2026.9 (2026-09-21) Blog Rank Why this rank top features scored 35/21/13 — 23 How ranking works →

Burp Suite 2026.9 adds the ability to export SMTPSMTPThe application-layer protocol used to send email between servers and from clients to servers; cyber tools use it to relay alerts, phishing simulations, or exfiltrated data over standard mail infrastructure. Collaborator interactions as email files, alongside expanded hotkey and title bar customization and command palette performance improvements.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

We enable the world to secure the web. Burp Suite, the Web Security Academy, and world-leading research from PortSwigger — the AppSec community’s home.

Burp Suite 2026.9 adds the ability to export SMTPSMTPThe application-layer protocol used to send email between servers and from clients to servers; cyber tools use it to relay alerts, phishing simulations, or exfiltrated data over standard mail infrastructure. Collaborator interactions as email files, alongside expanded hotkey and title bar customization and command palette performance improvements.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 3 FEATURES 0 OF 3 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
SMTP Collaborator interaction export new 35

Adds the ability to export SMTP Collaborator interactions as email files.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

— Names the feature but no export mechanism or location detail Professional / Community 2026.9 (2026-09-21)
02 00
Hotkey and title bar customization improved 21

Expands hotkey and title bar customization options.

— Vague on which options were added Professional / Community 2026.9 (2026-09-21)
03 00
Command palette performance improvements improved 13

Improves command palette performance.

— No detail on scope or magnitude of improvement Professional / Community 2026.9 (2026-09-21)
Professional / Community 2026.9 (2026-09-21) Blog LatestStable

Burp Suite 2026.9 adds SMTPSMTPThe application-layer protocol used to send email between servers and from clients to servers; cyber tools use it to relay alerts, phishing simulations, or exfiltrated data over standard mail infrastructure. Collaborator email export and expands hotkey and title bar customization.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Adds the ability to export SMTP Collaborator interactions as email files.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Expands hotkey and title bar customization options.
  • Improves command palette performance.
  • Adds the ability to export SMTP Collaborator interactions as email files.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

2 more in Everything

Was this useful?
Reverse Engineering, Malware & Forensics
◆  Malware Analysis

CAPEv2

Sources Commits → changes since 2026-09-09 Code

Malware Configuration And Payload Extraction

CAPEv2 adds agent_worktree.py for isolated PR/branch checkouts and SHA256SHA256A cryptographic hash function that turns data into a fixed 256-bit fingerprint. It is used to check integrity; hashing does not encrypt the original data. deduplication in ResultServer⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ TRY IT
Check out a contributor's PR into an isolated worktree so CI bots or reviewers can test it without risking uncommitted work in the main clone.
$ python utils/agent_worktree.py new --pr 3219
List all managed worktrees with their status flags (main / managed / dirty / pr#N), then tear down all managed ones after testing is complete.
$ python utils/agent_worktree.py list
python utils/agent_worktree.py cleanup --all
  • Adds utils/agent_worktree.py with subcommands new --pr, new --branch, new --from, list, path, update, info, remove, and cleanup --all for managing disposable git worktree checkouts of PRs or branches without touching the working clone; every subcommand accepts --repo PATH and --json flags.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • ResultServer now performs SHA256 deduplication and file versioning instead of raising CuckooOperationalError / FileExistsError when an uploaded filename already exists in storage/analyses/<task>/files/.
  • Adds utils/agent_worktree.py with subcommands new --pr, new --branch, new --from, list, path, update, info, remove, and cleanup --all for managing disposable git worktree checkouts of PRs or branches without touching the working clone; every subcommand accepts --repo PATH and --json flags.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

1 more in Everything

Was this useful?
◆  Reverse Engineering

pwndbg

Sources Release notes →Source code → 2026.09.15 Notes Code Rank Why this rank top features scored 80/70/60, +15 breaking change, +10 product media, +6 runnable example — 101 How ranking works →

Exploit Development and Reverse Engineering with GDB & LLDB Made Easy

pwndbg's 2026.09.15 release adds a new stack-vis command, extends nearpc to disassemble whole functions without debug symbols, introduces backwards disassembly at breakpoints, expands page-info commands, enables x64 kernel emulation, and improves handling of segfaulting indirect jumps — alongside a breaking rename in the debugger error API.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 7 FEATURES 1 OF 7 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Symbol-free function disassembly in nearpc improved 80

Expands nearpc -f/--function to work without debugging symbols, making it a full replacement for GDB's/LLDB's disass command with branch visualization, branch targets, and operand annotations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Disassemble an entire function by address without debug symbols, getting branch visualization and operand annotations that raw disass lacks.
$ nearpc --function 0x401080
View command output 2026.09.15 Real pwndbg output from a synthetic AArch64 core fixture.
● actually ran exit 0 · 1,118ms · b50806fc385a

GDB 15.1; inner_function, 0x401080–0x4010e0. Fixture function deliberately placed at 0x401080. No DWARF debug information; ordinary ELF symbols retained.

Fixture and debugger provenance 
{
  "source_sha": "55596aa676a2756efb8d67034f31675170eabe48",
  "source_commit": null,
  "image_digest": "sha256:b50806fc385a239dbc36fb053d6db9bea21c15881e0d9adda37d94200c353d81",
  "captured_at": "2026-09-23T00:49:08.684392+00:00",
  "execution_context": {
    "adapter": {
      "id": "pwndbg-core-aarch64-v1",
      "sha256": "ea642ca005adc4451df1a10b5ca103b86792619798b4e17d8f2e4659b2b46389"
    },
    "binding_sha256": "c5d0c09ebd28e3dfc9ca2756f158d7f847a8f12d4eb7c6d89ddc954cadb203ce",
    "debugger": {
      "argv": [
        "gdb",
        "-q",
        "-nx",
        "-nh",
        "-batch",
        "-ex",
        "set pagination off",
        "-ex",
        "set confirm off",
        "-ex",
        "set debuginfod enabled off",
        "-ex",
        "source /pwndbg/gdbinit.py",
        "-ex",
        "set debug-file-directory /nonexistent",
        "-ex",
        "file /opt/ctk/pwndbg/fixture.elf",
        "-ex",
        "core-file /opt/ctk/pwndbg/fixture.core",
        "-x",
        "/opt/ctk/pwndbg/capture.py"
      ],
      "command": "nearpc --function 0x401080",
      "core_loaded": true,
      "environment": {
        "CTK_SELECTED_DEBUGGER_COMMAND": "nearpc --function 0x401080",
        "DEBUGINFOD_URLS": "",
        "HOME": "/tmp",
        "LC_ALL": "C.UTF-8",
        "PWNDBG_NO_AUTOUPDATE": "1",
        "PWNDBG_VENV_PATH": "/venv",
        "PYTHONDONTWRITEBYTECODE": "1",
        "TERM": "dumb"
      },
      "frame": "inner_function",
      "name": "gdb",
      "parent_frame": "outer_function",
      "readable_stack_bytes": 64,
      "version": "15.1",
      "working_directory": "/tmp"
    },
    "fixture": {
      "architecture": "aarch64",
      "c_sha256": "1fa72d6195fe63c465c60822c7ca00bb3243afe80ae58086e0f786b2938f83d6",
      "compiler": {
        "argv": [
          "gcc",
          "-g",
          "-O0",
          "-nostdlib",
          "-no-pie",
          "-fno-pie",
          "-fno-stack-protector",
          "-fno-omit-frame-pointer",
          "-fno-optimize-sibling-calls",
          "-Wl,--section-start=.canary_target=0x401080,-e,_start",
          "-o",
          "/opt/ctk/pwndbg/fixture.elf",
          "/opt/ctk/pwndbg/fixture.c"
        ],
        "name": "gcc",
        "version": "13.3.0"
      },
      "core_sha256": "37cf63a26a5f5958a4800a6a3706e85cfb8efe7ab7c3b083583d547673beb89d",
      "debug_info": "absent",
      "elf_sha256": "234f99f28388e33d559114a758a0ffe3addc10c245b86999c426989322b028d8",
      "external_debug_links": false,
      "function": {
        "end": 4198624,
        "name": "inner_function",
        "start": 4198528
      },
      "kind": "synthetic_source_derived",
      "label": "Synthetic AArch64 core fixture",
      "source_files": {
        "tests/binaries/host/crash_simple.native.c": "b0387aff0ad0ff2258b03ced1808dff665a68565d73e790a75b16be9ed1dbe5d",
        "tests/binaries/host/stack_vars.native.c": "e911a591fae1d76c5ec5049f9ae6d50b01c0e3b25422844997de0e61526cb67d"
      },
      "source_repo": "pwndbg/pwndbg",
      "source_sha": "55596aa676a2756efb8d67034f31675170eabe48",
      "strip_argv": [
        "strip",
        "--strip-debug",
        "/opt/ctk/pwndbg/fixture.elf"
      ],
      "symbols": "elf_symtab"
    },
    "kind": "gdb_core",
    "output": {
      "scope": "selected_debugger_command",
      "sha256": "d2217eb166721f3949e98f3a1c546070382f1734b36c37fbb7036d13d738a9a0"
    }
  }
}
pwndbg · GDB 
      (gdb) nearpc --function 0x401080
   0x401080 <inner_function>                            sub    sp, sp, #0x50
   0x401084 <inner_function+4>                          adrp   x0, 0x400000        X0 => 0x400000 ◂— 0x10102464c457f
   0x401088 <inner_function+8>                          add    x1, x0, #0x1c0
   0x40108c <inner_function+12>                         add    x0, sp, #0x10
   0x401090 <inner_function+16>                         ldr    q0, [x1]
   0x401094 <inner_function+20>                         str    q0, [x0]
   0x401098 <inner_function+24>                         ldur   q0, [x1, #0xb]
   0x40109c <inner_function+28>                         stur   q0, [x0, #0xb]
   0x4010a0 <inner_function+32>                         add    x0, sp, #0x2b
   0x4010a4 <inner_function+36>                         movi   v0.4s, #0
   0x4010a8 <inner_function+40>                         stp    q0, q0, [x0]
   0x4010ac <inner_function+44>                         fmov   x1, d0
   0x4010b0 <inner_function+48>                         stur   x1, [x0, #0x1d]
   0x4010b4 <inner_function+52>                         mov    w0, #0x2a           W0 => 0x2a
   0x4010b8 <inner_function+56>                         str    w0, [sp, #0xc]
   0x4010bc <inner_function+60>                         ldr    w1, [sp, #0xc]
   0x4010c0 <inner_function+64>                         add    x0, sp, #0x10
   0x4010c4 <inner_function+68>                         cmp    w1, #0x2a
   0x4010c8 <inner_function+72>                    ┌<   b.eq   inner_function+80           <inner_function+80>
                                                   │ 
   0x4010cc <inner_function+76>                    │    nop   
 ► 0x4010d0 <inner_function+80>                    └>   udf    #0
   0x4010d4 <inner_function+84>                         nop   
   0x4010d8 <inner_function+88>                         add    sp, sp, #0x50
   0x4010dc <inner_function+92>                         ret
— Names flags and comparison target, runnable example given 2026.09.15
02 00
New stack-vis command for stack frames new 70

Adds the stack-vis command to visualize stack frames of the current thread, similar to how vis-heap-chunks visualizes heap chunks.

Visualize the current thread's stack frames at a breakpoint to understand frame layout during exploit development.
$ stack-vis
View command output 2026.09.15 Real pwndbg output from a synthetic AArch64 core fixture.
● actually ran exit 0 · 1,165ms · 0ee5b8fbd708

GDB 15.1; inner_function, 0x401080–0x4010e0. Fixture function deliberately placed at 0x401080. No DWARF debug information; ordinary ELF symbols retained.

Fixture and debugger provenance 
{
  "source_sha": "55596aa676a2756efb8d67034f31675170eabe48",
  "source_commit": null,
  "image_digest": "sha256:0ee5b8fbd708acf4a61c7ae73d418e74857cd28721e710444e466150c08cbd01",
  "captured_at": "2026-09-23T00:46:17.325163+00:00",
  "execution_context": {
    "adapter": {
      "id": "pwndbg-core-aarch64-v1",
      "sha256": "ea642ca005adc4451df1a10b5ca103b86792619798b4e17d8f2e4659b2b46389"
    },
    "binding_sha256": "ce3c84998475d87a23d05e9b38b490497a60c7c48f7a4a9a3e5ca9d066e7526d",
    "debugger": {
      "argv": [
        "gdb",
        "-q",
        "-nx",
        "-nh",
        "-batch",
        "-ex",
        "set pagination off",
        "-ex",
        "set confirm off",
        "-ex",
        "set debuginfod enabled off",
        "-ex",
        "source /pwndbg/gdbinit.py",
        "-ex",
        "set debug-file-directory /nonexistent",
        "-ex",
        "file /opt/ctk/pwndbg/fixture.elf",
        "-ex",
        "core-file /opt/ctk/pwndbg/fixture.core",
        "-x",
        "/opt/ctk/pwndbg/capture.py"
      ],
      "command": "stack-vis",
      "core_loaded": true,
      "environment": {
        "CTK_SELECTED_DEBUGGER_COMMAND": "stack-vis",
        "DEBUGINFOD_URLS": "",
        "HOME": "/tmp",
        "LC_ALL": "C.UTF-8",
        "PWNDBG_NO_AUTOUPDATE": "1",
        "PWNDBG_VENV_PATH": "/venv",
        "PYTHONDONTWRITEBYTECODE": "1",
        "TERM": "dumb"
      },
      "frame": "inner_function",
      "name": "gdb",
      "parent_frame": "outer_function",
      "readable_stack_bytes": 64,
      "version": "15.1",
      "working_directory": "/tmp"
    },
    "fixture": {
      "architecture": "aarch64",
      "c_sha256": "1fa72d6195fe63c465c60822c7ca00bb3243afe80ae58086e0f786b2938f83d6",
      "compiler": {
        "argv": [
          "gcc",
          "-g",
          "-O0",
          "-nostdlib",
          "-no-pie",
          "-fno-pie",
          "-fno-stack-protector",
          "-fno-omit-frame-pointer",
          "-fno-optimize-sibling-calls",
          "-Wl,--section-start=.canary_target=0x401080,-e,_start",
          "-o",
          "/opt/ctk/pwndbg/fixture.elf",
          "/opt/ctk/pwndbg/fixture.c"
        ],
        "name": "gcc",
        "version": "13.3.0"
      },
      "core_sha256": "085612a3b0edf5771376e9e1a7b19b4037463158b86a55685246f15c9d1c44d6",
      "debug_info": "absent",
      "elf_sha256": "234f99f28388e33d559114a758a0ffe3addc10c245b86999c426989322b028d8",
      "external_debug_links": false,
      "function": {
        "end": 4198624,
        "name": "inner_function",
        "start": 4198528
      },
      "kind": "synthetic_source_derived",
      "label": "Synthetic AArch64 core fixture",
      "source_files": {
        "tests/binaries/host/crash_simple.native.c": "b0387aff0ad0ff2258b03ced1808dff665a68565d73e790a75b16be9ed1dbe5d",
        "tests/binaries/host/stack_vars.native.c": "e911a591fae1d76c5ec5049f9ae6d50b01c0e3b25422844997de0e61526cb67d"
      },
      "source_repo": "pwndbg/pwndbg",
      "source_sha": "55596aa676a2756efb8d67034f31675170eabe48",
      "strip_argv": [
        "strip",
        "--strip-debug",
        "/opt/ctk/pwndbg/fixture.elf"
      ],
      "symbols": "elf_symtab"
    },
    "kind": "gdb_core",
    "output": {
      "scope": "selected_debugger_command",
      "sha256": "6c0d12dd3fd134959ab272cfd8af4d17c894b8ae9a4e4904d83f913e95409825"
    }
  }
}
pwndbg · GDB 
      (gdb) stack-vis
0xfffff3edba10	0x0000000000000000	0x0000002a00000000	............*...	 <-- inner_function+80
0xfffff3edba20	0x7266206f6c6c6548	0x72656e6e69206d6f	Hello from inner
0xfffff3edba30	0x6f6974636e756620	0x000000000000216e	 function!......
0xfffff3edba40	0x0000000000000000	0x0000000000000000	................
0xfffff3edba50	0x0000000000000000	0x0000000000000000	................
0xfffff3edba60	0x0000fffff3edba80	0x00000000004001b8	..........@.....	 <-- outer_function+32
0xfffff3edba70	0x0000000000000000	0x4142434445464748	........HGFEDCBA
0xfffff3edba80	0x0000000000000000	0x0000000000000000	................	 <-- _start+12
— Names command and analog, includes runnable example 2026.09.15
03 00
Full page info in v2p, p2v, pageinfo improved 60

Expands the v2p, p2v, and pageinfo commands to display full page info including vaddr, paddr, and page.

— Names commands and fields but no example given 2026.09.15
04 00
Backwards disassembly in context and nearpc new 60

Adds backwards disassembly in the context disasm view and nearpc, showing instructions before $PC when stopping at a breakpoint or after a syscall.

— Describes trigger conditions but lacks a usage example 2026.09.15
05 01
pwndbg.dbg_mod.Error renamed to DebuggerError breaking 60

pwndbg.dbg_mod.Error is renamed to pwndbg.dbg_mod.DebuggerError; any plugin or script that catches or references pwndbg.dbg_mod.Error will break on upgrade.

— Names exact symbol change and migration impact 2026.09.15
thinner coverage below
06 00
Improved indirect jump handling for x86 improved 55

Improves indirect jump handling for x86 PLT, ret instructions, and memory-indirect jumps that can segfault, enabling visualization of segfaulting ROP chains.

— Names affected instruction types, no direct command shown 2026.09.15
07 00
Emulation for x64 kernel debugging new 50

Enables emulation when debugging x64 kernels, unlocking emulated instruction flow and concrete operand display in the disassembly system.

— Explains benefit but no command or config surface named 2026.09.15
└──▷ ALSO FROM THESE RELEASES
└──▷ BREAKING ON UPGRADE
  • !
    pwndbg.dbg_mod.Error is renamed to pwndbg.dbg_mod.DebuggerError; any plugin or script that catches or references pwndbg.dbg_mod.Error will break on upgrade.
2026.09.15 Notes Code LatestStable

pwndbg gains backwards disassembly, new stack-vis command, nearpc -f without symbols, and full page info on v2p/p2v/pageinfo⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch 2026.09.15 https://github.com/pwndbg/pwndbg.git
# already have the repo? check out this version:
$ git checkout 2026.09.15
└──▷ TRY IT
Visualize the current thread's stack frames at a breakpoint to understand frame layout during exploit development.
$ stack-vis
Disassemble an entire function by address without debug symbols, getting branch visualization and operand annotations that raw disass lacks.
$ nearpc --function 0x401080
  • Adds stack-vis command to visualize stack frames of the current thread, similar to how vis-heap-chunks visualizes heap chunks.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Expands nearpc -f/--function to work without debugging symbols, making it a full replacement for GDB's/LLDB's disass command with branch visualization, branch targets, and operand annotations.
  • Expands v2p, p2v, and pageinfo commands to display full page info including vaddr, paddr, and page.
  • Adds backwards disassembly in context disasm and nearpc, showing instructions before $PC when stopping at a breakpoint or after a syscall.
  • Enables emulation when debugging x64 kernels, unlocking emulated instruction flow and concrete operand display in the disassembly system.
+1 more show less
  • Improves indirect jump handling for x86 PLT, ret instructions, and memory-indirect jumps that can segfault, enabling visualization of segfaulting ROP chains.
  • Adds stack-vis command to visualize stack frames of the current thread, similar to how vis-heap-chunks visualizes heap chunks.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds backwards disassembly in context disasm and nearpc, showing instructions before $PC when stopping at a breakpoint or after a syscall.

4 more in Everything

└──▷ BREAKING ON UPGRADE
  • !pwndbg.dbg_mod.Error is renamed to pwndbg.dbg_mod.DebuggerError; any plugin or script that catches or references pwndbg.dbg_mod.Error will break on upgrade.
Was this useful?
Network Security
◆  Network Analysis & Monitoring

Netcap

Sources Release notes →Source code → v0.9.1 Notes Code Rank Why this rank top features scored 85/83/78, +6 runnable example — 88 How ranking works →

Netcap v0.9.1 adds an MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. server for LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code.-driven PCAPPCAPA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic. analysis, structured ModbusModbusA serial communication protocol originally developed by Modicon in 1979 for connecting programmable logic controllers, now widely used in industrial control systems; cyber tools target it to monitor or interact with OT environments. and S7comm OT/ICSICSIndustrial Control Systems — hardware and software that monitor and control physical processes like power grids, pipelines, and manufacturing lines, making them a target for cyber tools focused on operational technology environments. threat-hunting with CISA-mapped rules, and Hyperscan-accelerated pattern matching, alongside service-mode, database, and web UI improvements.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

A framework for secure and scalable network traffic analysis - https://netcap.io

Netcap v0.9.1 adds an MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. server for LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code.-driven PCAPPCAPA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic. analysis, structured ModbusModbusA serial communication protocol originally developed by Modicon in 1979 for connecting programmable logic controllers, now widely used in industrial control systems; cyber tools target it to monitor or interact with OT environments. and S7comm OT/ICSICSIndustrial Control Systems — hardware and software that monitor and control physical processes like power grids, pipelines, and manufacturing lines, making them a target for cyber tools focused on operational technology environments. threat-hunting with CISA-mapped rules, and Hyperscan-accelerated pattern matching, alongside service-mode, database, and web UI improvements.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 11 FEATURES 0 OF 11 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
MCP server for LLM-driven PCAP analysis new 85

Adds a Model Context ProtocolModel Context ProtocolModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. (MCP) server exposing tools including query_audit_records, get_chart_data, extract_iocs_from_file, get_image_exif, lookup_cve, resolve, lookup_dhcp_fingerprint, get_live_capture_status, stop_live_capture, delete_session, and wait_for_session.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ lookup_cve fetches real NVDNVDThe National Vulnerability Database, maintained by NIST, is the U.S. government repository of CVE records enriched with CVSS scores and metadata; cyber tools query it to drive vulnerability prioritization. vulnerability data with caching, and the server exposes a netcap://session/{id}/graph resource, a Prometheus metrics endpoint at /mcp/metrics, and ?sessionId/?inputFile query parameters for per-request session selection shared with the web UI.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

— Names every tool and endpoint but no runnable example given v0.9.1
02 00
S7comm and S7CommPlus OT threat hunting new 83

Adds OT/ICS threat-hunting rules for Siemens S7commSiemens S7commA Siemens industrial protocol used by SIMATIC S7 PLCs to exchange programming, diagnostic, and process data, giving cyber tools a way to inspect or interact with Siemens automation networks. mapped to CISA advisory AA26-231A, covering write, logic download, logic theft, PLC stop, and CPUCPUThe main general-purpose processor in a computer. It runs application instructions and coordinates work with memory and other hardware. restart function codes, shipped as rules/examples/s7comm_hunt.yml. Also adds S7CommPlus cleartext framing parsing and a PayloadObscured flag to the S7comm stream decoder.

Hunt for unauthorized S7comm operations (logic download, PLC stop) on an OT network PCAP, with an allowlist of known-good engineering workstations to suppress false positives.
$ ./net capture -read ot-traffic.pcap --rules rules/examples/s7comm_hunt.yml --approved-workstations 10.0.1.50,10.0.1.51
— Named rule file, CISA mapping, and a runnable hunt command v0.9.1
03 00
Hyperscan/Vectorscan acceleration new 78

Adds an optional Hyperscan/Vectorscan prefilter for nmap service probes (~2.2x faster) and CMSCMSA web content management system for creating and managing digital content without writing code directly; cyber tools use CMS platforms to host documentation, portals, or reporting dashboards quickly./web framework detection (~1.4x), plus a Hyperscan-backed rule-engine MatchesPattern operator (up to ~6x faster on miss-heavy traffic); enabled by building with -tags hyperscan.

— Concrete speedups and build tag given, no full command v0.9.1
04 00
Structured Modbus decoding and RTU support new 75

Adds --modbus-rtu-endpoints (env NC_MODBUS_RTU_ENDPOINTS) to allowlist Modbus RTU-over-TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it. destinations (IP:port or [IPv6IPv6The current version of the Internet Protocol, standardized by the IETF, using 128-bit addresses to replace the exhausted IPv4 space. Cyber tools target it because modern networks run dual-stack or IPv6-only.]:port) for stream decoding, plus structured Modbus PDU decoding with threat-hunting rules covering all remaining Modbus function codes, pairing reads with writes via ordered correlation.

— Flag and env var named but no worked example shown v0.9.1
05 00
Service mode preload and database rebuild controls new 75

Adds --service-preload-dir (env NC_SERVICE_PRELOAD_DIR, default <service-data-dir>/pcaps) to preload a directory of PCAPsPCAPsA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic. when running in service mode, NC_DBS_SKIP_SOURCES to skip specific data sources during database rebuild, and a /health readiness endpoint on the database (dbs) HTTP listener available before the initial rebuild completes.

Start service mode and preload a directory of historical PCAPs for immediate LLM-driven analysis via the MCP server.
$ ./net capture --service --service-preload-dir /data/forensics/pcaps
— Flags, env vars and endpoint named with a runnable example v0.9.1
06 00
Time and behavior helpers in detection rule engine new 65

Adds source-to-distinct-destination cardinality (fan-out) detection and record-time thresholds to the detection rule engine, along with IsBusinessHours and HourOfDay time-of-day expression helpers and an IsApprovedWorkstation helper backed by a new --approved-workstations allowlist flag.

— Helpers and flag named but no dedicated example v0.9.1
thinner coverage below
07 00
Dashboard and fingerprints API in web UI new 55

Adds a configurable dashboard landing page with built-in security views to the web UI, and server-side filter, sort, and pagination plus a per-row community-ID cap to the /api/fingerprints endpoint.

— Endpoint named but no request example or UI path v0.9.1
08 00
Worker-owned TCP pools for concurrent capture improved 45

Replaces the global assembly lock with per-worker TCP pool ownership, enabling concurrent capture.

— Describes the mechanism but gives no config or flag v0.9.1
09 00
Deterministic offline capture output improved 35

Offline capture output is now deterministic: the same PCAP produces identical audit records across runs.

— States the before/after but no mechanism detail v0.9.1
10 00
GeoIP enrichment for Connection records new 25

Adds GeoIPGeoIPA database that maps IP addresses to geographic locations; tools use it to attribute traffic or alerts to countries, cities, or ASNs without querying external services. enrichment to Connection records.

— Single-line description with no mechanism or fields named v0.9.1
11 00
Fuzz targets for packet and stream parsers new 20

Adds 19 fuzz targets for packet and stream parsers.

— Bare count with no target names or usage v0.9.1
v0.9.1 Notes Code LatestStable

Netcap v0.9.1 adds an MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. server for LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code.-driven PCAPPCAPA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic. analysis, ModbusModbusA serial communication protocol originally developed by Modicon in 1979 for connecting programmable logic controllers, now widely used in industrial control systems; cyber tools target it to monitor or interact with OT environments. threat-hunting rules, S7comm OT detection, and Hyperscan acceleration.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch v0.9.1 https://github.com/dreadl0ck/netcap.git
# already have the repo? check out this version:
$ git checkout v0.9.1
└──▷ TRY IT
Hunt for unauthorized S7comm operations (logic download, PLC stop) on an OT network PCAP, with an allowlist of known-good engineering workstations to suppress false positives.
$ ./net capture -read ot-traffic.pcap --rules rules/examples/s7comm_hunt.yml --approved-workstations 10.0.1.50,10.0.1.51
Start service mode and preload a directory of historical PCAPsPCAPsA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic. for immediate LLM-driven analysis via the MCP server.
$ ./net capture --service --service-preload-dir /data/forensics/pcaps
  • Adds --modbus-rtu-endpoints flag (env: NC_MODBUS_RTU_ENDPOINTS) to specify a comma-separated allowlist of RTU-over-TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it. destinations (IP:port or [IPv6IPv6The current version of the Internet Protocol, standardized by the IETF, using 128-bit addresses to replace the exhausted IPv4 space. Cyber tools target it because modern networks run dual-stack or IPv6-only.]:port) for Modbus RTU stream decoding.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds --service-preload-dir flag (env: NC_SERVICE_PRELOAD_DIR) to specify a directory of PCAPs to preload when running in service mode (defaults to <service-data-dir>/pcaps).
  • Adds --approved-workstations flag for loading an approved-workstation allowlist used by detection rules via the IsApprovedWorkstation expression helper.
  • Adds an MCP (Model Context ProtocolModel Context ProtocolModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface.) server for LLM-driven PCAP analysis, exposing tools including query_audit_records, get_chart_data, extract_iocs_from_file, get_image_exif, lookup_cve, resolve, lookup_dhcp_fingerprint, get_live_capture_status, stop_live_capture, delete_session, and wait_for_session.
  • Adds lookup_cve MCP tool that fetches real NVDNVDThe National Vulnerability Database, maintained by NIST, is the U.S. government repository of CVE records enriched with CVSS scores and metadata; cyber tools query it to drive vulnerability prioritization. vulnerability data with caching.
+17 more show less
  • Adds netcap://session/{id}/graph MCP resource for session graph access.
  • Adds Prometheus metrics endpoint at /mcp/metrics for the MCP server.
  • Adds ?sessionId / ?inputFile query parameters for per-request session selection across the web UI and MCP server.
  • Adds source-to-distinct-destination cardinality (fan-out) detection and record-time thresholds to the detection rule engine.
  • Adds IsBusinessHours and HourOfDay time-of-day helpers to the detection rule expression engine.
  • Adds GeoIPGeoIPA database that maps IP addresses to geographic locations; tools use it to attribute traffic or alerts to countries, cities, or ASNs without querying external services. enrichment to Connection records.
  • Adds OT/ICSICSIndustrial Control Systems — hardware and software that monitor and control physical processes like power grids, pipelines, and manufacturing lines, making them a target for cyber tools focused on operational technology environments. threat-hunting rules for Siemens S7commSiemens S7commA Siemens industrial protocol used by SIMATIC S7 PLCs to exchange programming, diagnostic, and process data, giving cyber tools a way to inspect or interact with Siemens automation networks. mapped to CISA AA26-231A, covering write, logic download, logic theft, PLC stop, and CPUCPUThe main general-purpose processor in a computer. It runs application instructions and coordinates work with memory and other hardware. restart function codes; ships rules/examples/s7comm_hunt.yml.
  • Adds S7CommPlus cleartext framing parsing and a PayloadObscured flag to the S7comm stream decoder.
  • Adds structured Modbus PDU decoding with threat-hunting rules covering all remaining Modbus function codes, pairing reads with writes via ordered correlation.
  • Adds optional Hyperscan/Vectorscan prefilter for nmap service probes (~2.2× faster), CMSCMSA web content management system for creating and managing digital content without writing code directly; cyber tools use CMS platforms to host documentation, portals, or reporting dashboards quickly./web framework detection (~1.4×), and the rule-engine MatchesPattern operator (up to ~6× on miss-heavy traffic); build with -tags hyperscan.
  • Adds NC_DBS_SKIP_SOURCES environment variable to skip specific data sources during database rebuild.
  • Adds a /health readiness endpoint to the database (dbs) HTTP listener, available before the initial rebuild completes.
  • Adds configurable dashboard landing page with built-in security views to the web UI.
  • Adds server-side filter, sort, and pagination plus a per-row community-ID cap to the /api/fingerprints endpoint.
  • Adds 19 fuzz targets for packet and stream parsers.
  • Adds worker-owned TCP pools, replacing the global assembly lock with per-worker pool ownership for concurrent capture.
  • Makes offline capture output deterministic: the same PCAP now produces identical audit records across runs.
  • Adds --modbus-rtu-endpoints flag (env: NC_MODBUS_RTU_ENDPOINTS) to specify a comma-separated allowlist of RTU-over-TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it. destinations (IP:port or [IPv6IPv6The current version of the Internet Protocol, standardized by the IETF, using 128-bit addresses to replace the exhausted IPv4 space. Cyber tools target it because modern networks run dual-stack or IPv6-only.]:port) for Modbus RTU stream decoding.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds --service-preload-dir flag (env: NC_SERVICE_PRELOAD_DIR) to specify a directory of PCAPs to preload when running in service mode (defaults to <service-data-dir>/pcaps).
  • Adds --approved-workstations flag for loading an approved-workstation allowlist used by detection rules via the IsApprovedWorkstation expression helper.
  • Adds an MCP (Model Context ProtocolModel Context ProtocolModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface.) server for LLM-driven PCAP analysis, exposing tools including query_audit_records, get_chart_data, extract_iocs_from_file, get_image_exif, lookup_cve, resolve, lookup_dhcp_fingerprint, get_live_capture_status, stop_live_capture, delete_session, and wait_for_session.
  • Adds lookup_cve MCP tool that fetches real NVDNVDThe National Vulnerability Database, maintained by NIST, is the U.S. government repository of CVE records enriched with CVSS scores and metadata; cyber tools query it to drive vulnerability prioritization. vulnerability data with caching.
+9 more show less
  • Adds netcap://session/{id}/graph MCP resource for session graph access.
  • Adds Prometheus metrics endpoint at /mcp/metrics for the MCP server.
  • Adds ?sessionId / ?inputFile query parameters for per-request session selection across the web UI and MCP server.
  • Adds IsBusinessHours and HourOfDay time-of-day helpers to the detection rule expression engine.
  • Adds OT/ICSICSIndustrial Control Systems — hardware and software that monitor and control physical processes like power grids, pipelines, and manufacturing lines, making them a target for cyber tools focused on operational technology environments. threat-hunting rules for Siemens S7commSiemens S7commA Siemens industrial protocol used by SIMATIC S7 PLCs to exchange programming, diagnostic, and process data, giving cyber tools a way to inspect or interact with Siemens automation networks. mapped to CISA AA26-231A, covering write, logic download, logic theft, PLC stop, and CPUCPUThe main general-purpose processor in a computer. It runs application instructions and coordinates work with memory and other hardware. restart function codes; ships rules/examples/s7comm_hunt.yml.
  • Adds optional Hyperscan/Vectorscan prefilter for nmap service probes (~2.2× faster), CMSCMSA web content management system for creating and managing digital content without writing code directly; cyber tools use CMS platforms to host documentation, portals, or reporting dashboards quickly./web framework detection (~1.4×), and the rule-engine MatchesPattern operator (up to ~6× on miss-heavy traffic); build with -tags hyperscan.
  • Adds NC_DBS_SKIP_SOURCES environment variable to skip specific data sources during database rebuild.
  • Adds a /health readiness endpoint to the database (dbs) HTTP listener, available before the initial rebuild completes.
  • Adds 19 fuzz targets for packet and stream parsers.

8 more in Everything

Was this useful?

NETRESEC

Sources Blog / feed → Unmasking Malware Families (2026-09-21) Blog Rank Why this rank top features scored 95/70, +6 runnable example — 88.5 How ranking works →

NetResec introduced FlowCarp, a statistical C2 protocol classifier that identifies malware families from packet behaviour rather than signatures, shipping both a CLI for training custom models and a hosted demo API.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Network forensics, packet sniffers and IT security products. Download NetworkMiner and other free software for network security analysis.

NetResec introduced FlowCarp, a statistical C2 protocol classifier that identifies malware families from packet behaviour rather than signatures, shipping both a CLI for training custom models and a hosted demo API.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 2 FEATURES 1 OF 2 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 01
FlowCarp CLI for statistical C2 protocol classification new 95

FlowCarp is a new statistical application-layer protocol classifier that identifies C2 protocols — including encrypted and proprietary ones — by analyzing packet sizes, timing, message length distributions, and request/response behavior rather than byte-sequence signatures.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ The flowcarp CLI adds --input, --output, and --model flags for building and applying custom protocol models from PCAPPCAPA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic. files: flowcarp --input <pcap> --output model:<label>:<class>:<file>.fcp trains a new model from example C2 capture sessions, and flowcarp --model <file>.fcp --input <pcap> --output - classifies traffic in a new PCAP against a previously built custom model.

Train a custom FlowCarp model from sandbox-captured C2 sessions so you can classify unknown malware traffic without writing IDSIDSAn intrusion detection system, monitoring traffic or host activity for suspicious behavior and producing alerts. Detection alone does not necessarily block the activity. signatures.
$ flowcarp --input xyz.pcap --output model:MalwareXYZ:MALICIOUS:malwarexyz.fcp
Classify traffic from a new sandbox capture against your custom protocol model to confirm C2 family attribution.
$ flowcarp --model malwarexyz.fcp --input new.pcap --output -
— Names exact flags, commands and file format for training/classifying Unmasking Malware Families (2026-09-21)
02 00
Hosted FlowCarp demo REST API new 70

A demo REST endpoint at https://demo.flowcarp.com accepts raw PCAP via HTTP POST and returns a JSON response of detected flows, protocols, and alerts. The hosted demo service detects more than 600 protocols and supports adding custom protocol models built from user-supplied example traffic.

Submit a PCAP to the FlowCarp demo service and pipe the JSON results into jq for quick triage during incident response.
$ curl --data-binary @dump.pcap https://demo.flowcarp.com | jq
— Endpoint and response format named but limited mechanism detail Unmasking Malware Families (2026-09-21)
Unmasking Malware Families (2026-09-21) Blog LatestStable

FlowCarp brings statistical C2C2Command-and-control infrastructure used by attackers to remotely direct compromised systems; cyber tools use it to simulate or detect adversary communication channels during red-team or threat-detection work. protocol classification to PCAPPCAPA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic. analysis, identifying 600+ malware families without signature rules.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ TRY IT
Train a custom FlowCarp model from sandbox-captured C2 sessions so you can classify unknown malware traffic without writing IDSIDSAn intrusion detection system, monitoring traffic or host activity for suspicious behavior and producing alerts. Detection alone does not necessarily block the activity. signatures.
$ flowcarp --input xyz.pcap --output model:MalwareXYZ:MALICIOUS:malwarexyz.fcp
Classify traffic from a new sandbox capture against your custom protocol model to confirm C2 family attribution.
$ flowcarp --model malwarexyz.fcp --input new.pcap --output -
Submit a PCAP to the FlowCarp demo service and pipe the JSON results into jq for quick triage during incident response.
$ curl --data-binary @dump.pcap https://demo.flowcarp.com | jq
  • Adds --input, --output, and --model flags to the flowcarp CLI for building and applying custom statistical protocol models from PCAP files.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds flowcarp --input <pcap> --output model:<label>:<class>:<file>.fcp to train a new protocol model from example C2 capture sessions.
  • Adds flowcarp --model <file>.fcp --input <pcap> --output - to classify traffic in a new PCAP against a previously built custom model.
  • Exposes a demo REST endpoint at https://demo.flowcarp.com accepting raw PCAP via HTTP POST and returning a JSON response of detected flows, protocols, and alerts.
  • Introduces FlowCarp, a statistical application-layer protocol classifier that identifies C2 protocols — including encrypted and proprietary ones — by analyzing packet sizes, timing, message length distributions, and request/response behavior rather than byte-sequence signatures.
+1 more show less
  • Supports detection of more than 600 protocols on the hosted demo service, with the ability to add custom protocol models from user-supplied example traffic.

6 details in Everything

Was this useful?
◆  Firewalls & Network Defense

Fastly

Sources Press release →Blog post → announced · Sep 21, 2026 Press Rank Why this rank top features scored 77/55/45, +10 product media — 69 How ranking works →

Fastly launched an AI-focused security and governance stack at the edge: AI Runtime Control (ARC) as a managed control plane for LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code. traffic, an AI Firewall to catch prompt injectionprompt injectionAn attack where text the model reads, such as a web page, file or tool result, carries instructions that override the user's. It is the main security risk for agents with tools. attacks, and API Security enforcement extended to agentic AI traffic.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Fastly's edge cloud platform delivers faster, safer, and more scalable sites and apps to customers. Elevate your edge CDN, video delivery, security, and more.

Fastly launched an AI-focused security and governance stack at the edge: AI Runtime Control (ARC) as a managed control plane for LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code. traffic, an AI Firewall to catch prompt injectionprompt injectionAn attack where text the model reads, such as a web page, file or tool result, carries instructions that override the user's. It is the main security risk for agents with tools. attacks, and API Security enforcement extended to agentic AI traffic.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 3 FEATURES 2 OF 3 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 01
AI Runtime Control (ARC) for LLM traffic new 77

ARC is a Fastly-managed control plane that routes every model call through a single endpoint across public and self-hosted AI providers, providing access governance, real-time token spend visibility, rate limiting, and resilience.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Per-app and per-developer virtual keys can be created, rotated, or revoked instantly without client-side code changes, replacing shared provider API keys and protecting underlying provider credentials; enforceable monthly dollar-budget limits can be applied directly to individual virtual keys. Automatic failover lets a virtual key point at an ordered list of model targets (primary, fallback, and further) so a provider outage is transparent to callers, and ARC delivers full request and response logging for every AI interaction (prompts and completions), searchable by user, model, provider, and date range. It operates as bring-your-own-key (BYOKBYOKBring Your Own Key — a cloud deployment model where the customer supplies and controls the encryption keys rather than the provider, so sensitive data stays under the customer's cryptographic authority.), preserving existing provider relationships and negotiated rates.

— Rich mechanism and named controls but no API/CLI syntax given launch-20260921-de13b213launch-20260921-0c070049
thinner coverage below
02 02
AI Firewall for prompt injection detection new 55

AI Firewall evaluates prompts directly in the request path at the edge to detect and block LLM-based attacks, including prompt injection, before they reach targeted models. It is powered by an updated version of Fastly's SmartParse detection engine and, within ARC, supports optional custom guardrails.

— Names detection engine but no config or endpoint given launch-20260921-de13b213launch-20260921-0c070049
03 00
API Security enforcement for agentic AI traffic new 45

Fastly's API Security now applies API contract rules to agentic, agent-assisted, and conventional traffic on a service-by-service basis, letting organizations observe or block non-conforming requests from AI agents accessing enterprise APIs.

— Describes scope but no concrete config or rule syntax launch-20260921-0c070049
announced · Sep 21, 2026 Press

Fastly launches AI Runtime Control, AI Firewall, and expanded API Security to govern, protect, and control AI traffic at the edge.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Adds AI Runtime Control, which routes model calls through a single endpoint across public and self-hosted AI providers, with virtual keys to protect underlying provider credentials, real-time token spend visibility, rate limiting, budget controls, and failover.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds AI Firewall, which evaluates prompts directly in the request path at the edge to detect and block LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code.-based attacks including prompt injectionprompt injectionAn attack where text the model reads, such as a web page, file or tool result, carries instructions that override the user's. It is the main security risk for agents with tools. before they reach targeted models.
  • Adds API Security enforcement for agentic traffic, applying API contract rules to agentic, agent-assisted, and conventional traffic service by service — allowing organizations to observe or block non-conforming requests from AI agents accessing enterprise APIs.

3 details in Everything

announced · Sep 21, 2026 Blog

Fastly launches AI Runtime Control (ARC), a control plane for governing, securing, and auditing all LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code. provider traffic.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Introduces AI Runtime Control (ARC), a Fastly-managed control plane that routes every LLM model call through a single endpoint for access governance, spend attribution, and resilience.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Provides per-app and per-developer virtual keys that can be created, rotated, or revoked instantly without code changes on the client side, replacing shared provider API keys.
  • Applies enforceable monthly dollar-budget limits directly to individual virtual keys for proactive AI spend control.
  • Supports automatic failover by letting a virtual key point at an ordered list of model targets (primary, fallback, and further) so a provider outage is transparent to callers.
  • Includes AI Firewall within ARC — powered by an updated version of Fastly's SmartParse detection engine — to detect prompt injectionprompt injectionAn attack where text the model reads, such as a web page, file or tool result, carries instructions that override the user's. It is the main security risk for agents with tools. and other LLM-specific attacks via optional custom guardrails.
+2 more show less
  • Delivers full request and response logging for every AI interaction (prompts and completions), searchable by user, model, provider, and date range, to support compliance and audit requirements.
  • Operates as bring-your-own-key (BYOKBYOKBring Your Own Key — a cloud deployment model where the customer supplies and controls the encryption keys rather than the provider, so sensitive data stays under the customer's cryptographic authority.), preserving existing provider relationships and negotiated rates while adding the governance and visibility layer around them.

7 details in Everything

Was this useful?
Cloud & Container Security
◆  Runtime Security

Falco

Sources Release notes → 0.45.0 Notes Rank Why this rank top features scored 90/85/75, +15 breaking change, +6 runnable example — 104.33 How ranking works →

Cloud Native Runtime Security

Falco 0.45 reworks rule matching to operate on raw bytes and adds new multi-value string modifiers, alongside a reload-status API endpoint and several smaller CLI and Helm chart additions.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 6 FEATURES 2 OF 6 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
GET /reload API and Unix socket reload control new 90

Exposes reload status via GET /reload and adds optional reload control over a protected Unix socket on Linux, preserving SIGHUP reception and pending requests across reloads.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Status can be checked with curl -s --unix-socket /run/falco/falco.sock http://localhost/reload.

Check whether Falco has finished reloading after sending SIGHUP, without restarting the process.
$ curl -s --unix-socket /run/falco/falco.sock http://localhost/reload
— Runnable curl command and exact endpoint path given 0.45.0
02 01
Raw-byte rule matching and regex sanitization breaking 85

Rule conditions now evaluate on raw field bytes, enabling matching of invalid and non-printable UTF-8 sequences with string operators and single-byte matching via the new \xHH escape sequence (e.g. evt.buffer contains "\x00"). The regex operator remains an exception: it operates on sanitized input, replacing invalid UTF-8 bytes with U+FFFD before matching, while non-printable characters are escaped at output encoding time. This is a breaking change: conditions using non-regex operators to match U+FFFD in place of invalid bytes will no longer behave the same and must be updated to match the original raw bytes; FALCO_ENGINE_VERSION is bumped from 0.62.0 to 0.63.0.

Catch filenames containing a raw null byte or other non-printable byte (e.g. evasion via NUL-injection) using the new \xHH hex escape in a byte-buffer condition.
text 
evt.buffer contains "\x00"
— Names exact escape syntax and version bump, but no migration script 0.45.0
03 00
anyof/allof/oneof string comparison modifiers new 75

Adds anyof, allof, and oneof string comparison modifiers (since Falco 0.44) so a single condition can test one field against multiple values or patterns, e.g. proc.cmdline contains anyof (curl, wget) or proc.name != allof (bash, sh). Also adds support for referencing a Falco list inside the modifier parentheses, e.g. fd.name contains anyof (suspicious_paths).

Detect processes invoking common exfiltration tools without writing a separate rule for each binary, using the anyof modifier to match multiple values in one condition.
text 
proc.cmdline contains anyof (curl, wget, nc, scp)
Exclude multiple known-good shell names in a single condition using allof with !=, rather than chaining separate inequality checks.
env 
proc.name != allof (bash, sh)
— Concrete syntax with runnable rule conditions given product docs
04 02
container_engines config key deprecation warning deprecated 60

Configurations still setting the removed container_engines key will now log a schema validation warning at startup, since the key has been dropped from the configuration JSON schema.

— Names the exact key removed but no replacement guidance 0.45.0
thinner coverage below
05 00
serviceAccount.labels in Helm chart new 50

Adds serviceAccount.labels to the Helm chart to set custom labels on the ServiceAccount.

— Names the config key but no further mechanism or example 0.45.0
06 00
--dry-run suggestion on --validate improved 40

Falco now suggests --dry-run when --validate receives a Falco configuration file.

— Names both flags but no example output shown 0.45.0
└──▷ BREAKING ON UPGRADE
  • !
    When upgrading from Falco 0.44 or earlier, conditions that used non-regex operators to match the Unicode replacement character U+FFFD in place of invalid bytes will no longer behave the same — those conditions must be updated to match the original raw bytes instead.
  • !
    Rules that matched the UTF-8 replacement character using non-regex operators must be reviewed: rule conditions now operate on raw field bytes, and FALCO_ENGINE_VERSION is bumped from 0.62.0 to 0.63.0.
  • !
    Configurations still setting the removed container_engines key will now log a schema validation warning at startup (the key has been dropped from the configuration JSON schema).
0.45.0 Notes LatestStable

Falco 0.45.0 adds a GET /reload API endpoint, raw-byte rule matching with \xHH escape sequences, and a --dry-run suggestion on --validate.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch 0.45.0 https://github.com/falcosecurity/falco.git
# already have the repo? check out this version:
$ git checkout 0.45.0
└──▷ TRY IT
Check whether Falco has finished reloading after sending SIGHUP, without restarting the process.
$ curl -s --unix-socket /run/falco/falco.sock http://localhost/reload
  • Exposes reload status via GET /reload and adds optional reload control over a protected Unix socket on Linux, preserving SIGHUP reception and pending requests across reloads.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Evaluates rule conditions on raw field bytes, enabling matching of invalid and non-printable UTF-8 sequences with string operators and single-byte matching with the new \xHH escape sequence; regex now operates on sanitized input and non-printable characters are escaped at output encoding time.
  • Adds serviceAccount.labels to the Helm chart to set custom labels on the ServiceAccount.
  • Suggests --dry-run when --validate receives a Falco configuration file.
  • Exposes reload status via GET /reload and adds optional reload control over a protected Unix socket on Linux, preserving SIGHUP reception and pending requests across reloads.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

3 more in Everything

└──▷ BREAKING ON UPGRADE
  • !Rules that matched the UTF-8 replacement character using non-regex operators must be reviewed: rule conditions now operate on raw field bytes, and FALCO_ENGINE_VERSION is bumped from 0.62.0 to 0.63.0.
  • !Configurations still setting the removed container_engines key will now log a schema validation warning at startup (the key has been dropped from the configuration JSON schema).
Was this useful?

Alibaba Cloud OpenSandbox

Sources Release notes → release-1.1.0 Notes Rank Why this rank top features scored 90/88/85, +15 breaking change, +6 runnable example — 108.67 How ranking works →

Secure, Fast, and Extensible Sandbox runtime for AI agents.

OpenSandbox 1.1.0 ships Firecracker FastPath fleets, an opt-in PostgreSQL snapshot store with HA coordination, sandbox lifecycle hooks, and a scoped diagnostics API, alongside a wide set of egress, security, and Kubernetes provider hardening changes — plus breaking migrations to agents.x-k8s.io/v1beta1agents.x-k8s.io/v1beta1A Kubernetes API group version for the CAPI (Cluster API) Agents provider, maintained under the x-k8s.io experimental namespace, letting cyber tools manage cluster infrastructure lifecycle via standard Kubernetes custom resources. and removal of the informer_enabled flag.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 17 FEATURES 3 OF 17 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Scoped diagnostics API new 90

Adds a stable, scoped diagnostics API with logs:container|all and events:runtime|lifecycle|all scopes, returning 400 DIAGNOSTICS_SCOPE_UNSUPPORTED for unknown scopes and backward-compatible plain-text responses.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Scope diagnostics queries to only runtime events for a sandbox, using the new events:runtime scope to cut noise during incident triage.
$ curl -H 'OPEN-SANDBOX-API-KEY: <key>' 'https://<server>/diagnostics?scope=events:runtime'
— Runnable curl example with exact scopes and error code release-1.1.0
02 00
PostgreSQL snapshot store with HA coordination new 88

Adds an opt-in PostgreSQL snapshot store with pooled connections and CAS semantics, a one-shot migrate-snapshots command for moving existing SQLite records, and multi-process HA coordination for snapshot create/recovery/delete on Kubernetes.

Enable the PostgreSQL snapshot store for HA multi-process deployments and migrate existing SQLite records in one shot.
$ migrate-snapshots
— Runnable migrate-snapshots command included release-1.1.0
03 00
Egress sidecar resource, OTLP export, and timeout config new 85

Adds [egress.resources] for independent resource requests/limits on the Kubernetes egress sidecar, [egress] otlp_endpoint so the sidecar exports metrics to an OpenTelemetryOpenTelemetryAn open standard and set of tools for collecting traces, metrics and logs from applications. It lets services export observability data without tying instrumentation to one monitoring vendor. collector on both runtimes, and [egress] timeout_seconds to make the sidecar's readiness wait configurable.

Export egress sidecar metrics to your OpenTelemetry collector and cap its resource footprint independently of the sandbox container.
toml 
[egress]
otlp_endpoint = "http://otel-collector:4317"
timeout_seconds = 30

[egress.resources]
requests = { cpu = "50m", memory = "64Mi" }
limits   = { cpu = "200m", memory = "128Mi" }
— Exact config keys with runnable TOML example release-1.1.0
04 00
Sandbox lifecycle hooks (OSEP-0020) new 80

Adds preStart and periodic lifecycle hooks to sandbox create requests, validated server-side and serialized into a reserved environment variable for Kubernetes workloads, with preStart.timeoutSeconds raised to 10800 seconds. Lifecycle hook models were added to the API spec and every SDK with validation consistent with server-side ranges.

— Names field, env var, and timeout limit but no example call release-1.1.0
05 00
Pool-only sandbox creation and back-pressure signaling improved 80

Supports pool-only sandbox creation using poolRef without an entrypoint (previously rejected with HTTP 400), and surfaces pool back-pressure as HTTP 429 KUBERNETES::POOL_CAPACITY_EXHAUSTED with a Retry-After header.

— Exact status codes and field named, no curl example release-1.1.0
06 01
agent-sandbox provider migrates to v1beta1 breaking 75

The agent-sandbox provider and ingress informer migrate to agents.x-k8s.io/v1beta1 (required by agent-sandbox v1.0.0), creating workloads with operatingMode: Running and service: true; existing workloads using the prior API version will break.

— Exact API version and fields named, no migration steps release-1.1.0
07 00
Docker runtime config parity with Kubernetes new 70

Adds sandbox_env and sandbox_binds Docker-runtime-wide configuration keys, giving Docker parity with the Kubernetes pod template for fleet-wide CA bundles, proxies, and mounts.

— Named config keys, no example command release-1.1.0
08 00
FastPath fleets lifecycle backend new 70

Adds the FastPath fleets lifecycle backend: create, renew, delete, and status mapping via FastPath v2, plus end-to-end fast-sandbox pause/resume with async 202 semantics and UID fencing.

— Mechanism named, no exact command shown release-1.1.0
09 00
Secure-access signing keys from env/Secret improved 65

Sources OSEP-0011 secure-access signing keys from the OPENSANDBOX_SECURE_ACCESS_KEYS environment variable or a Kubernetes Secret instead of plaintext chart values.

— Named env var, no walkthrough of migration release-1.1.0
10 00
Per-sandbox image pull secrets new 65

Honors image.auth in create requests for the agent-sandbox provider, with per-sandbox pull secrets garbage-collected along with the Sandbox CR.

— Named field and GC behavior, no example release-1.1.0
11 00
Configurable proxy internal resolution new 60

Adds [proxy] resolve_internal config key to make the reverse proxy's internal resolution configurable for deployments that cannot route to Docker bridge IPs.

— Named config key but no example given release-1.1.0
12 00
gVisor snapshot rejection new 60

Rejects unsupported gVisorgVisorA user-space kernel written in Go by Google that intercepts and handles system calls on behalf of a container, sandboxing it without needing a full VM. (runsc) snapshots before persistence with 409 SNAPSHOT::UNSUPPORTED_RUNTIME.

— Exact error code named, no reproduction example release-1.1.0
thinner coverage below
13 00
Server-side HTTP request metrics over OTLP new 55

Exports generic HTTP request metrics (request rate, error rate, latency) from the server over OTLPOTLPOpenTelemetry Protocol, a standard way to send traces, metrics and logs between applications, collectors and monitoring systems. It lets telemetry move across compatible tools., complementing the SDK-reported create-duration histogram.

— Named metrics and OTLP but no endpoint/config key release-1.1.0
14 02
Strict CodeInterpreter health check by default breaking 55

Defaults CodeInterpreter creation to a strict health check in every SDK language, blocking until both execd and the Jupyter kernel gateway are verified ready.

— Explains new default behavior, no opt-out flag given release-1.1.0
15 00
Egress telemetry spoofing prevention improved 50

Injects OPENSANDBOX_EGRESS_SANDBOX_ID server-side so egress sidecar telemetry can no longer be spoofed by request environment variables.

— Explains fix but no reproduction steps release-1.1.0
16 03
informer_enabled flag removed breaking 50

The informer_enabled flag is removed; the informer-backed cache is now the only Kubernetes read path and the flag can no longer be used to disable it.

— Named flag and removal, no replacement guidance release-1.1.0
17 00
FastPath snapshot and template management in SDKs new 40

Adds fast-sandbox snapshot support and fsb template management across all five SDKs.

— Thin description, no SDK method names given release-1.1.0
└──▷ BREAKING ON UPGRADE
  • !
    The agent-sandbox provider and ingress informer migrate to agents.x-k8s.io/v1beta1 (required by agent-sandbox v1.0.0), creating workloads with operatingMode: Running and service: true; existing workloads using the prior API version will break.
  • !
    The informer_enabled flag is removed; the informer-backed cache is now the only Kubernetes read path and the flag can no longer be used to disable it.
release-1.1.0 Notes LatestStable

OpenSandbox 1.1.0 debuts Firecracker FastPath sandboxes, a PostgreSQL snapshot store, lifecycle hooks, and a scoped diagnostics API⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch release-1.1.0 https://github.com/opensandbox-group/OpenSandbox.git
# already have the repo? check out this version:
$ git checkout release-1.1.0
└──▷ TRY IT
Enable the PostgreSQL snapshot store for HA multi-process deployments and migrate existing SQLite records in one shot.
$ migrate-snapshots
Scope diagnostics queries to only runtime events for a sandbox, using the new events:runtime scope to cut noise during incident triage.
$ curl -H 'OPEN-SANDBOX-API-KEY: <key>' 'https://<server>/diagnostics?scope=events:runtime'
Export egress sidecar metrics to your OpenTelemetryOpenTelemetryAn open standard and set of tools for collecting traces, metrics and logs from applications. It lets services export observability data without tying instrumentation to one monitoring vendor. collector and cap its resource footprint independently of the sandbox container.
toml 
[egress]
otlp_endpoint = "http://otel-collector:4317"
timeout_seconds = 30

[egress.resources]
requests = { cpu = "50m", memory = "64Mi" }
limits   = { cpu = "200m", memory = "128Mi" }
  • Adds preStart and periodic lifecycle hooks to sandbox create requests (OSEP-0020), validated server-side and serialized into a reserved environment variable for Kubernetes workloads; preStart.timeoutSeconds limit raised to 10800 seconds.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds opt-in PostgreSQL snapshot store with pooled connections and CAS semantics, plus the one-shot migrate-snapshots command for moving existing SQLite records, and multi-process HA coordination for snapshot create/recovery/delete on Kubernetes.
  • Adds [egress.resources] config key for independent resource requests/limits on the Kubernetes egress sidecar.
  • Adds [egress] otlp_endpoint config key so the egress sidecar exports metrics to an OpenTelemetry collector on both runtimes.
  • Adds [egress] timeout_seconds config key to make egress sidecar readiness wait configurable.
+13 more show less
  • Adds [proxy] resolve_internal config key to make the reverse proxy's internal resolution configurable for deployments that cannot route to Docker bridge IPs.
  • Adds sandbox_env and sandbox_binds Docker-runtime-wide configuration keys, giving Docker parity with the Kubernetes pod template for fleet-wide CA bundles, proxies, and mounts.
  • Sources OSEP-0011 secure-access signing keys from the OPENSANDBOX_SECURE_ACCESS_KEYS environment variable or a Kubernetes Secret instead of plaintext chart values.
  • Injects OPENSANDBOX_EGRESS_SANDBOX_ID server-side so egress sidecar telemetry can no longer be spoofed by request environment variables.
  • Adds a stable, scoped diagnostics API with logs:container|all and events:runtime|lifecycle|all scopes, returning 400 DIAGNOSTICS_SCOPE_UNSUPPORTED for unknown scopes and backward-compatible plain-text responses.
  • Adds the FastPath fleets lifecycle backend: create, renew, delete, and status mapping via FastPath v2, plus end-to-end fast-sandbox pause/resume with async 202 semantics and UID fencing.
  • Adds fast-sandbox snapshot support and fsb template management across all five SDKs.
  • Exports generic HTTP request metrics (request rate, error rate, latency) from the server over OTLPOTLPOpenTelemetry Protocol, a standard way to send traces, metrics and logs between applications, collectors and monitoring systems. It lets telemetry move across compatible tools., complementing the SDK-reported create-duration histogram.
  • Rejects unsupported gVisorgVisorA user-space kernel written in Go by Google that intercepts and handles system calls on behalf of a container, sandboxing it without needing a full VM. (runsc) snapshots before persistence with 409 SNAPSHOT::UNSUPPORTED_RUNTIME.
  • Supports pool-only sandbox creation using poolRef without an entrypoint (previously rejected with HTTP 400), and surfaces pool back-pressure as HTTP 429 KUBERNETES::POOL_CAPACITY_EXHAUSTED with a Retry-After header.
  • Honors image.auth in create requests for the agent-sandbox provider with per-sandbox pull secrets garbage-collected with the Sandbox CR.
  • Adds sandbox lifecycle hook models (OSEP-0020) to the API spec and every SDK with consistent validation aligned to server ranges.
  • Defaults CodeInterpreter creation to a strict health check in every SDK language, blocking until both execd and the Jupyter kernel gateway are verified ready.
  • Adds preStart and periodic lifecycle hooks to sandbox create requests (OSEP-0020), validated server-side and serialized into a reserved environment variable for Kubernetes workloads; preStart.timeoutSeconds limit raised to 10800 seconds.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds opt-in PostgreSQL snapshot store with pooled connections and CAS semantics, plus the one-shot migrate-snapshots command for moving existing SQLite records, and multi-process HA coordination for snapshot create/recovery/delete on Kubernetes.
  • Adds [egress.resources] config key for independent resource requests/limits on the Kubernetes egress sidecar.
  • Adds [egress] otlp_endpoint config key so the egress sidecar exports metrics to an OpenTelemetry collector on both runtimes.
  • Adds [egress] timeout_seconds config key to make egress sidecar readiness wait configurable.
+10 more show less
  • Adds [proxy] resolve_internal config key to make the reverse proxy's internal resolution configurable for deployments that cannot route to Docker bridge IPs.
  • Adds sandbox_env and sandbox_binds Docker-runtime-wide configuration keys, giving Docker parity with the Kubernetes pod template for fleet-wide CA bundles, proxies, and mounts.
  • Injects OPENSANDBOX_EGRESS_SANDBOX_ID server-side so egress sidecar telemetry can no longer be spoofed by request environment variables.
  • Adds a stable, scoped diagnostics API with logs:container|all and events:runtime|lifecycle|all scopes, returning 400 DIAGNOSTICS_SCOPE_UNSUPPORTED for unknown scopes and backward-compatible plain-text responses.
  • Adds the FastPath fleets lifecycle backend: create, renew, delete, and status mapping via FastPath v2, plus end-to-end fast-sandbox pause/resume with async 202 semantics and UID fencing.
  • Adds fast-sandbox snapshot support and fsb template management across all five SDKs.
  • Exports generic HTTP request metrics (request rate, error rate, latency) from the server over OTLPOTLPOpenTelemetry Protocol, a standard way to send traces, metrics and logs between applications, collectors and monitoring systems. It lets telemetry move across compatible tools., complementing the SDK-reported create-duration histogram.
  • Rejects unsupported gVisorgVisorA user-space kernel written in Go by Google that intercepts and handles system calls on behalf of a container, sandboxing it without needing a full VM. (runsc) snapshots before persistence with 409 SNAPSHOT::UNSUPPORTED_RUNTIME.
  • Honors image.auth in create requests for the agent-sandbox provider with per-sandbox pull secrets garbage-collected with the Sandbox CR.
  • Adds sandbox lifecycle hook models (OSEP-0020) to the API spec and every SDK with consistent validation aligned to server ranges.

3 more in Everything

└──▷ BREAKING ON UPGRADE
  • !The agent-sandbox provider and ingress informer migrate to agents.x-k8s.io/v1beta1agents.x-k8s.io/v1beta1A Kubernetes API group version for the CAPI (Cluster API) Agents provider, maintained under the x-k8s.io experimental namespace, letting cyber tools manage cluster infrastructure lifecycle via standard Kubernetes custom resources. (required by agent-sandbox v1.0.0), creating workloads with operatingMode: Running and service: true; existing workloads using the prior API version will break.
  • !The informer_enabled flag is removed; the informer-backed cache is now the only Kubernetes read path and the flag can no longer be used to disable it.
Was this useful?
◆  CNAPP

Plerion

Sources Release page → API keys now hold a role: Notes Rank Why this rank top features scored 88/81/80, +6 runnable example — 89 How ranking works →

Plerion introduced role-based access controlrole-based access controlRole-based access control, where permissions are assigned to roles such as reader or administrator and users receive those roles. It makes access easier to manage than individual permission lists. for tenant API keys, including a role-scoped OpenAPIOpenAPIA vendor-neutral specification for describing REST APIs in a machine-readable format, maintained by the OpenAPI Initiative. Cyber tools use it to auto-generate clients, documentation, and validation from a single source of truth. discovery endpoint and a new permission gate on IaCIaCInfrastructure as Code: the practice of defining and provisioning compute infrastructure through machine-readable config files rather than manual processes, giving cyber tools repeatable, version-controlled environment setup. scan uploads, and shipped five new or expanded AWS security checks covering Redshift VPCVPCA logically isolated virtual network inside a public cloud provider's infrastructure, letting tools deploy resources with private addressing, controlled routing, and firewall rules without managing physical hardware. routing, Security Hub AI best practices, Kinesis encryption, Organizations exit controls, and ALBALBAWS Application Load Balancer, a managed Layer 7 load balancer from Amazon that routes HTTP/HTTPS traffic by host, path, or header, giving tools flexible ingress control without managing their own proxy. access logging.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Plerion is a cloud security platform that discovers and manages vulnerabilities and misconfigurations across infrastructure.

Plerion introduced role-based access controlrole-based access controlRole-based access control, where permissions are assigned to roles such as reader or administrator and users receive those roles. It makes access easier to manage than individual permission lists. for tenant API keys, including a role-scoped OpenAPIOpenAPIA vendor-neutral specification for describing REST APIs in a machine-readable format, maintained by the OpenAPI Initiative. Cyber tools use it to auto-generate clients, documentation, and validation from a single source of truth. discovery endpoint and a new permission gate on IaCIaCInfrastructure as Code: the practice of defining and provisioning compute infrastructure through machine-readable config files rather than manual processes, giving cyber tools repeatable, version-controlled environment setup. scan uploads, and shipped five new or expanded AWS security checks covering Redshift VPCVPCA logically isolated virtual network inside a public cloud provider's infrastructure, letting tools deploy resources with private addressing, controlled routing, and firewall rules without managing physical hardware. routing, Security Hub AI best practices, Kinesis encryption, Organizations exit controls, and ALBALBAWS Application Load Balancer, a managed Layer 7 load balancer from Amazon that routes HTTP/HTTPS traffic by host, path, or header, giving tools flexible ingress control without managing their own proxy. access logging.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 8 FEATURES 1 OF 8 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 01
Permission gate on IaC scan uploads breaking 88

IaC scan uploads are now gated behind a new CodeRepository:RequestScan permission on POST /v1/tenant/shiftleft/iac/scan; custom roles can add this permission via Request code scans under Manage integrations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Upload an IaC scan result using a key whose role carries the CodeRepository:RequestScan permission.
$ curl -s -X POST https://api.plerion.com/v1/tenant/shiftleft/iac/scan \
  -H 'Authorization: Bearer <your-api-key>' \
  -H 'Content-Type: application/json' \
  -d @scan-results.json
— Named endpoint, permission and UI path with curl example API keys now hold a role:
02 00
Role-scoped OpenAPI reference endpoint new 81

New GET /v1/tenant/openapi endpoint returns the API reference scoped to the calling key's role, letting a key discover exactly what it is permitted to call.

Audit what API operations a specific key is allowed to call — useful when scoping a least-privilege integration key.
$ curl -s -H 'Authorization: Bearer <your-api-key>' https://api.plerion.com/v1/tenant/openapi
— Named endpoint with a runnable curl example API keys now hold a role:
03 00
Redshift enhanced VPC routing check new 80

New check PLERION-AWS-329 (High severity) fails any Redshift cluster whose EnhancedVpcRouting setting is off, ensuring COPY and UNLOAD traffic to S3 stays inside the VPC and is visible to security groups, network ACLsACLsAn access control list, a set of rules allowing or denying operations for particular users, services or network addresses. It decides who can use a resource., endpoint policies, and VPC Flow Logs. Maps to AWS Foundational Security Best Practices Redshift.7, CISCISA nonprofit that publishes consensus-based configuration benchmarks and controls for hardening operating systems, cloud environments, and software; cyber tools use it as a ready-made compliance baseline. Controls v8, and NIST 800-53NIST 800-53A catalog of security and privacy controls for U.S. federal information systems, published by NIST. Cyber tools cite it to map their findings or features to a recognized compliance framework..

— Named check ID and framework mappings, no remediation command Say hello to a new Redshift check:
04 00
Security Hub AI Best Practices standard check new 80

New check PLERION-AWS-1177 (Medium severity) fails any AWS region where the Security Hub AI Security Best Practices standard is not in READY state, covering 31 automated controls for Amazon Bedrock, Amazon Bedrock AgentCore, and Amazon SageMaker. Maps to the AWS Well-Architected Machine Learning Lens and Generative AI Lens.

— Named check ID and scope, no remediation steps given Say hello to a new Redshift check:
05 00
Kinesis Data Streams encryption check new 73

New check PLERION-AWS-1176 fails any Kinesis Data Streams channel encrypted with an AWS-owned key rather than a customer-managed KMSKMSA managed service that creates, stores, rotates and controls access to encryption keys, usually backed by HSMs. Applications ask it to encrypt or decrypt rather than holding keys themselves. key; Plerion now collects Kinesis channels as a new asset type.

— Named check ID and new asset type, no remediation command Say hello to a new Redshift check:
06 00
Organization member-account exit control check new 73

New check PLERION-AWS-1178 (High severity) fails any AWS organization where at least one active member account is not covered by an unconditional SCPSCPSecure Copy Protocol, a file-transfer method built on SSH that encrypts data in transit; cyber tools use it to move files to or from remote hosts without exposing credentials or content. Deny of both organizations:LeaveOrganization and account:CloseAccount.

— Named check ID and exact SCP actions required Say hello to a new Redshift check:
07 00
Role assignment for API keys new 65

Tenant API keys now carry a role: new keys can be created with Tenant read-only or a custom role, and an existing key's role can be changed from the API keys list in the console.

Assign a least-privilege role to an API key at creation time so it cannot perform write operations.
📍In the Plerion console, go to API keys › Create key, set Role to Tenant read-only, then save. To update an existing key, select it from the API keys list and choose a different role.
— UI steps given but no API endpoint for the update itself API keys now hold a role:
08 00
ALB access logging check now recognizes CloudWatch Logs improved 62

Check PLERION-AWS-791 ('Application Load Balancer does not have access logging enabled') now passes ALBs that deliver access logs to CloudWatch Logs in addition to those logging to S3.

— Named check ID and before/after but no remediation steps Say hello to a new Redshift check:
API keys now hold a role: Notes Stable

Plerion API keys now carry a role, and a new IaCIaCInfrastructure as Code: the practice of defining and provisioning compute infrastructure through machine-readable config files rather than manual processes, giving cyber tools repeatable, version-controlled environment setup. scan upload permission gates POST /v1/tenant/shiftleft/iac/scan.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ TRY IT
Audit what API operations a specific key is allowed to call — useful when scoping a least-privilege integration key.
$ curl -s -H 'Authorization: Bearer <your-api-key>' https://api.plerion.com/v1/tenant/openapi
Upload an IaC scan result using a key whose role carries the CodeRepository:RequestScan permission.
$ curl -s -X POST https://api.plerion.com/v1/tenant/shiftleft/iac/scan \
  -H 'Authorization: Bearer <your-api-key>' \
  -H 'Content-Type: application/json' \
  -d @scan-results.json
Assign a least-privilege role to an API key at creation time so it cannot perform write operations.
📍In the Plerion console, go to API keys › Create key, set Role to Tenant read-only, then save. To update an existing key, select it from the API keys list and choose a different role.
  • Adds role assignment to every tenant API key; new keys can be created with Tenant read-only or a custom role, and an existing key's role can be changed from the API keys list.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds GET /v1/tenant/openapi endpoint that returns the API reference scoped to the calling key's role, so a key can discover exactly what it is permitted to call.
  • Gates IaC scan uploads behind a new CodeRepository:RequestScan permission on POST /v1/tenant/shiftleft/iac/scan; custom roles can add it via Request code scans under Manage integrations.

3 details in Everything

Say hello to a new Redshift check: Notes Stable

Plerion adds four new checks: Redshift enhanced VPCVPCA logically isolated virtual network inside a public cloud provider's infrastructure, letting tools deploy resources with private addressing, controlled routing, and firewall rules without managing physical hardware. routing, Security Hub AI Best Practices, Kinesis channel encryption, and Organizations member-account exit controls.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Adds check PLERION-AWS-329 (High severity) that fails any Redshift cluster whose EnhancedVpcRouting setting is off, ensuring COPY and UNLOAD traffic to S3 stays inside the VPC and is visible to security groups, network ACLsACLsAn access control list, a set of rules allowing or denying operations for particular users, services or network addresses. It decides who can use a resource., endpoint policies, and VPC Flow Logs; maps to AWS Foundational Security Best Practices Redshift.7, CISCISA nonprofit that publishes consensus-based configuration benchmarks and controls for hardening operating systems, cloud environments, and software; cyber tools use it as a ready-made compliance baseline. Controls v8, and NIST 800-53NIST 800-53A catalog of security and privacy controls for U.S. federal information systems, published by NIST. Cyber tools cite it to map their findings or features to a recognized compliance framework..⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds check PLERION-AWS-1177 (Medium severity) that fails any AWS region where the Security Hub AI Security Best Practices standard is not in READY state, covering 31 automated controls for Amazon Bedrock, Amazon Bedrock AgentCore, and Amazon SageMaker; maps to the AWS Well-Architected Machine Learning Lens and Generative AI Lens.
  • Adds check PLERION-AWS-1176 that fails any Kinesis Data Streams channel encrypted with an AWS-owned key rather than a customer-managed KMSKMSA managed service that creates, stores, rotates and controls access to encryption keys, usually backed by HSMs. Applications ask it to encrypt or decrypt rather than holding keys themselves. key; Plerion now collects Kinesis channels as a new asset type.
  • Adds check PLERION-AWS-1178 (High severity) that fails any AWS organization where at least one active member account is not covered by an unconditional SCPSCPSecure Copy Protocol, a file-transfer method built on SSH that encrypts data in transit; cyber tools use it to move files to or from remote hosts without exposing credentials or content. Deny of both organizations:LeaveOrganization and account:CloseAccount.
  • Expands check PLERION-AWS-791 ('Application Load Balancer does not have access logging enabled') to pass ALBs that deliver access logs to CloudWatch Logs in addition to those logging to S3.
  • Adds check PLERION-AWS-329 (High severity) that fails any Redshift cluster whose EnhancedVpcRouting setting is off, ensuring COPY and UNLOAD traffic to S3 stays inside the VPC and is visible to security groups, network ACLsACLsAn access control list, a set of rules allowing or denying operations for particular users, services or network addresses. It decides who can use a resource., endpoint policies, and VPC Flow Logs; maps to AWS Foundational Security Best Practices Redshift.7, CISCISA nonprofit that publishes consensus-based configuration benchmarks and controls for hardening operating systems, cloud environments, and software; cyber tools use it as a ready-made compliance baseline. Controls v8, and NIST 800-53NIST 800-53A catalog of security and privacy controls for U.S. federal information systems, published by NIST. Cyber tools cite it to map their findings or features to a recognized compliance framework..⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds check PLERION-AWS-1177 (Medium severity) that fails any AWS region where the Security Hub AI Security Best Practices standard is not in READY state, covering 31 automated controls for Amazon Bedrock, Amazon Bedrock AgentCore, and Amazon SageMaker; maps to the AWS Well-Architected Machine Learning Lens and Generative AI Lens.
  • Adds check PLERION-AWS-1176 that fails any Kinesis Data Streams channel encrypted with an AWS-owned key rather than a customer-managed KMSKMSA managed service that creates, stores, rotates and controls access to encryption keys, usually backed by HSMs. Applications ask it to encrypt or decrypt rather than holding keys themselves. key; Plerion now collects Kinesis channels as a new asset type.
  • Adds check PLERION-AWS-1178 (High severity) that fails any AWS organization where at least one active member account is not covered by an unconditional SCPSCPSecure Copy Protocol, a file-transfer method built on SSH that encrypts data in transit; cyber tools use it to move files to or from remote hosts without exposing credentials or content. Deny of both organizations:LeaveOrganization and account:CloseAccount.

1 more in Everything

Was this useful?
Detection & Response
◆  Endpoint Security (EDR/XDR)

Sophos Mobile

Sources Release page → 2026.38Released on September 21, 2026 Notes Rank Why this rank top features scored 60/53/40 — 51 How ranking works →

Sophos Mobile 2026.38 adds automatic MTDMTDMoving Target Defense: a strategy that continuously shifts a system's attack surface — rotating IPs, ports, or configurations — so attackers cannot rely on static reconnaissance data. policy hardening for API-created policies, new Apple Business profile setup-assistant skip options, and support for the latest Apple OS versions.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Sophos Mobile is an enterprise mobile device management and security solution that protects smartphones and tablets through threat detection, compliance, and remote management.

Sophos Mobile 2026.38 adds automatic MTDMTDMoving Target Defense: a strategy that continuously shifts a system's attack surface — rotating IPs, ports, or configurations — so attackers cannot rely on static reconnaissance data. policy hardening for API-created policies, new Apple Business profile setup-assistant skip options, and support for the latest Apple OS versions.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 3 FEATURES 0 OF 3 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
New Apple Business profile setup-assistant settings new 60

Apple Business profiles gain three new settings to skip corresponding setup-assistant steps: Accessibility Appearance, Safety & Handling, and Liquid Glass.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ The Accessibility Appearance and Safety & Handling settings are iPhone/iPad only.

— Names exact settings but no walkthrough of use 2026.38Released on September 21, 2026
thinner coverage below
02 00
Auto-config for Mobile API-created MTD policies improved 53

Sophos Mobile now automatically adds recommended configurations to Mobile Threat Defense policies created via the Mobile API. This applies only to newly created policies; existing policies are left unchanged.

— Explains scope and before/after but no config keys given 2026.38Released on September 21, 2026
03 00
iOS 27, iPadOS 27, and macOS 27 support new 40

Sophos Mobile adds support for iOS 27, iPadOS 27, and macOS 27 as managed device platforms.

— Bare platform support announcement with no detail on new capabilities 2026.38Released on September 21, 2026
2026.38Released on September 21, 2026 Notes LatestStable

Sophos Mobile 2026.38 adds iOS/iPadOS/macOS 27 support, new Apple Business profile settings, and auto-config for Mobile API-created MTDMTDMoving Target Defense: a strategy that continuously shifts a system's attack surface — rotating IPs, ports, or configurations — so attackers cannot rely on static reconnaissance data. policies.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Automatically adds recommended configurations to Mobile Threat Defense policies created via the Mobile API (applies to new policies only; existing policies are unchanged).⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds Accessibility Appearance, Safety & Handling, and Liquid Glass settings to Apple Business profiles to skip the corresponding setup-assistant steps (Accessibility Appearance and Safety & Handling are iPhone/iPad only).
  • Adds support for iOS 27, iPadOS 27, and macOS 27 as managed device platforms.
  • Automatically adds recommended configurations to Mobile Threat Defense policies created via the Mobile API (applies to new policies only; existing policies are unchanged).⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

2 more in Everything

Was this useful?

Trend Vision One

Sources Blog / feed → Local response filters now officially released (2026-09-21) Blog Rank Why this rank top features scored 80/35, +6 runnable example — 63.5 How ranking works →

Trend Vision One reached general availability for local response filters that auto-terminate processes on Windows endpoints, and the Zero Trust Secure Access On-Premises Gateway added content inspection for Vertex Chat traffic.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Trend Vision One is Trend Micro's cloud-based XDR platform providing unified security visibility and threat detection across endpoints, networks, and cloud environments.

Trend Vision One reached general availability for local response filters that auto-terminate processes on Windows endpoints, and the Zero Trust Secure Access On-Premises Gateway added content inspection for Vertex Chat traffic.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 2 FEATURES 1 OF 2 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 01
Local response filters for Windows endpoints new 80

Local response filters are now generally available, automatically terminating processes on Windows endpoints upon detection to reduce mean time to detect.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ They can be managed via Endpoint Security → Endpoint Security Configuration → Endpoint Security Policies, activated per-detection from Agentic SIEM and XDR → Observed Attack Techniques, and all enabled filters can be listed under Agentic SIEM and XDR → Detection Model Management → Local response filters.

Audit which local response filters are currently active across your environment before enabling new ones.
📍In the console, go to Agentic SIEM and XDR → Detection Model Management → Local response filters to view all enabled filters.
Embed automated process-termination responses into a Windows endpoint policy so detections are actioned without manual intervention.
📍In the console, go to Endpoint Security → Endpoint Security Configuration → Endpoint Security Policies, select the target policy, and enable the desired local response filters.
Trigger automated process termination directly from an observed attack technique event to shorten response time for a specific detection.
📍In the console, go to Agentic SIEM and XDR → Observed Attack Techniques, locate the relevant technique event, and enable the associated local response filter.
— Exact console navigation paths given for each workflow Local response filters now officially releas…
thinner coverage below
02 00
Vertex Chat content inspection in ZTSA gateway new 35

The Zero Trust Secure Access On-Premises Gateway (version 1.0.94) adds content inspection support for Vertex Chat traffic passing through the gateway.

— Names the traffic type but no mechanism or steps Zero Trust Secure Access On-Premises Gateway…
Local response filters now officially released (2026-09-21) Blog Stable

Trend Vision One GA: local response filters auto-terminate processes on Windows endpoints to cut MTTD.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Audit which local response filters are currently active across your environment before enabling new ones.
📍In the console, go to Agentic SIEM and XDR → Detection Model Management → Local response filters to view all enabled filters.
Embed automated process-termination responses into a Windows endpoint policy so detections are actioned without manual intervention.
📍In the console, go to Endpoint Security → Endpoint Security Configuration → Endpoint Security Policies, select the target policy, and enable the desired local response filters.
Trigger automated process termination directly from an observed attack technique event to shorten response time for a specific detection.
📍In the console, go to Agentic SIEM and XDR → Observed Attack Techniques, locate the relevant technique event, and enable the associated local response filter.
  • Enables local response filters for Windows endpoints that automatically terminate processes on detection, reducing mean time to detect.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Manages local response filters via Endpoint Security → Endpoint Security Configuration → Endpoint Security Policies to integrate detection into security policy workflows.
  • Activates local response filters per-detection in Agentic SIEM and XDR → Observed Attack Techniques to convert individual event detections into automated responses.
  • Lists all enabled local response filters in Agentic SIEM and XDR → Detection Model Management → Local response filters.

4 details in Everything

Zero Trust Secure Access On-Premises Gateway version 1.0.94 (2026-09-21) Blog LatestStable

Zero Trust Secure Access On-Premises Gateway 1.0.94 adds Vertex Chat content inspection support.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Was this useful?
Identity, Access & Data Protection
◆  Identity & Access Management

JumpCloud

Sources Blog / feed → Release Notes 2022 (2026-09-21) Blog Rank Why this rank top features scored 90/90/85, +15 breaking change, +10 product media, +6 runnable example — 119.33 How ranking works →

JumpCloud makes it simple to manage Windows, Apple, Linux, and Android devices and to deliver secure access with SSO, MFA, Zero Trust, RADIUS, and more.

Across releases spanning 2022–2026, JumpCloud shipped its broadest identity and device-management expansion yet: Password Vault across mobile, Admin Portal, User Portal and browser extension; Password Manager and Remote Assist maturing into full-featured tools; JumpCloud Go passwordless auth, MCP ServerMCP ServerA process that exposes tools, resources, and prompts over Anthropic's Model Context Protocol, letting an AI agent call external APIs or local services through a standardised interface., AI Commands Builder and Agent Identities; GA launches of Access Risk Detection and SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally. License Management; and breaking changes to the systemusers reactivate endpoint, Dynamic Groups filter operators, and LDAPLDAPA protocol for querying and modifying directory services, standardized under RFC 4511, commonly used to look up users, groups, and attributes in corporate identity stores like Active Directory. TLS cipher suites.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 64 FEATURES 10 OF 64 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Systemusers reactivate endpoint replaced breaking 90

Replaces the deprecated GET api/systemusers/reactivate/{id} endpoint, which now returns a 404, with POST https://console.jumpcloud.com/systemusers/reactivate/{id} accepting activation_email in the request body; all callers must switch to the new endpoint.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Reactivate a user account after migrating scripts from the deprecated GET endpoint to the new POST endpoint.
$ POST https://console.jumpcloud.com/systemusers/reactivate/{id}
{
  "activation_email": "{email_address}"
}
— Exact old and new endpoints plus a runnable request body example Release Notes 2024 (2026-09-21)
02 01
Agent Identities for non-human AI agents new 90

Introduces Agent Identities (Public Preview) under Identity Management > Agents to register and govern non-human AI agent accounts with OAuth-based authentication, group-based application access, and full lifecycle management independent of human user accounts. Adds a client authentication type setting on an agent's Authentication tab supporting Client secret basic (credentials in the authorization header, default) and Client secret post (credentials in the request body) for OAuth token requests.

Register a new AI agent and scope its OAuth credentials to require credentials in the request body — needed when integrating with platforms like Amazon Bedrock AgentCore that enforce client_secret_post.
📍In the Admin Portal: Go to Identity Management › Agents › + Add. Fill in agent name, owner, and company email on the Details tab. On the Authentication tab, generate OAuth client credentials, then set the client authentication type to Client secret post to send the client ID and secret in the request body.
— Named UI path, both OAuth modes, and a full setup example Release Notes 2026 (2026-09-21)
03 00
AI Gateway MCP visibility and dashboard new 85

Adds an MCP Connection column (showing Discovered, Connected, or Not Available) in Access > AI & SaaS Management > Applications to surface MCP server status and add servers to AI Gateway without switching tools, plus a Dashboard tab on AI Gateway with org-wide MCP server usage metrics, tool call volume, error rates, trend charts, and per-server drill-down.

Surface which SaaS apps have locally running MCP servers and connect them to AI Gateway for centralized governance.
📍In the Admin Portal: Go to Access › AI & SaaS Management › Applications. Review the MCP Connection column for apps showing Discovered. Click the status for a target app, review discovered MCP servers, then click Add MCP Server to register them in AI Gateway.
— Exact column values and navigation path with a worked example Release Notes 2026 (2026-09-21)
04 02
Password Vault across Admin Portal, User Portal, mobile app, and browser extension new 83

Launches Password Vault as a password-management product spanning four surfaces: the JumpCloud Protect mobile app (view secrets, autofill into mobile apps and mobile browser using Push/TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip.-authenticated sessions), the Admin Portal (enable the vault, assign access to users and user groups, manage shared credentials and folders with Connect/View/Edit/Manage permissions, and view audit logs in Directory Insights), the User Portal (manage and share credentials and website applications, bulk import via CSV or Excel, and organize items in personal and shared folders), and a Browser Extension via JumpCloud Go (autofill of non-SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. website passwords and 2FA codes, credential saving on login/update/signup, and secret viewing without requiring device registration).

— Names each surface and permission model but no exact menu path Release Notes 2026 (2026-09-21)
05 00
Dynamic Groups filter operators renamed breaking 81

Replaces deprecated $le and $ge Dynamic Groups memberQuery filter operators with $lte and $gte to align with the rest of the API list endpoints; existing API scripts using $le or $ge must be updated (Admin Portal-managed Dynamic Groups are unaffected).

— Exact deprecated and replacement operator names given Release Notes 2024 (2026-09-21)
06 00
Step-up authentication and MFA proof in SSO new 80

Adds a sso_auth log event specifically for forced re-authentication events in SAMLSAMLA standard for passing signed identity assertions between a sign-in provider and an application. It lets a user sign in through their organization's identity provider./OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. flows for compliance visibility, ForceAuthn=true (SAML) and prompt=login (OIDC) support so applications can trigger step-up authentication without ending the user's existing session, and Authentication Methods References (AMRAMRAutonomous Mobile Robot — a self-navigating wheeled platform that maps its environment and plans its own path; cyber tools might target AMR firmware or fleet-management APIs in OT/ICS assessments.) in SAML assertions and OIDC ID tokens to supply MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. proof to connected applications.

Audit which authentication factors were used during an SSO login by inspecting the AMR claim in a JumpCloud-issued OIDC ID token — useful for verifying Zero Trust policy compliance.
📍In the Admin Portal, go to Directory Insights and filter by event type sso_auth to view forced re-authentication events and confirm AMR tags (e.g., password + OTP) for each login.
— Names exact SAML/OIDC parameters and log event with an audit example Release Notes 2025 (2026-09-21)
07 03
LDAP security hardening breaking 80

Restricts JumpCloud LDAP TLS to five modern cipher suites — TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256, TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256, TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384, TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384, and TLS_RSA_WITH_AES_128_GCM_SHA256 — removing CBC-mode and SHA-1SHA-1A cryptographic hash function standardized by NIST that produces a 160-bit digest; tools use it for file integrity checks, though it is now considered cryptographically broken and unsuitable for collision-resistant applications./SHA-256SHA-256A cryptographic hash function that turns data into a fixed 256-bit fingerprint. It is used to check integrity; hashing does not encrypt the original data. suites; any LDAP client configured to use only removed suites will fail to connect. Also adds source IP tracking for incoming LDAP requests for enhanced security and traceability.

— Names every allowed cipher suite and the connection-failure impact Release Notes 2026 (2026-09-21)Release Notes 2025 (2026-09-21)
08 00
Password Manager desktop app capacity, sharing, and session controls improved 78

Adds an Auto Lock setting that locks the app automatically on device lock or after a configurable auto-lock schedule (default 15 minutes of inactivity, up to 1440 minutes); raises item storage to up to 13,000 items with autofill enabled, 5,000 items per shared folder, and 30,000 items without autofill; introduces a paginated items list showing 100 items per page; adds a Move All items button for shared folders and the ability to delete multiple folders at once; adds JumpCloud Password Manager Cloud Backups as a secure fallback to minimize vault data loss; lets admins create and update shared folders and assign users or user groups to any folder, building on earlier Shared Folders management with per-folder manual sync; adds item archiving (view, delete, restore), multi-select and sorting by nickname, a Copy button in Password History, syncing of shared folder events to Directory Insights, and a What's New link plus Download Logs and Keyboard Shortcuts options under Settings.

— Names exact settings and limits but no CLI/API commands What’s New: Password Manager Desktop App (20…Release Notes 2023 (2026-09-21)
09 04
JumpCloud Go passwordless authentication new 78

Adds the JumpCloud Go browser extension for hardware-protected, phishing-resistant passwordless login on managed macOS and Windows using Apple Touch ID and Windows HelloWindows HelloA Microsoft authentication framework built into Windows 10 and later that lets users sign in via biometrics or PIN instead of passwords, giving cyber tools a phishing-resistant credential layer to build on., set as the default MFA method for conditional access policies on SSO apps with user and device verification on every session. Adds AWS Verified Access integration supporting JumpCloud Go, SAML, and OIDC SSO as a VPNVPNA virtual private network, an encrypted tunnel between devices or networks. It carries traffic across another network as though the endpoints had a private connection.-less remote access route to corporate web resources, a User Portal > Security > Multi-Factor Authentication section for managing Go sessions, and automatic session refresh when a user logs in to or unlocks their managed device.

— Names the extension, integration and portal path but no CLI command Release Notes 2024 (2026-09-21)Release Notes 2023 (2026-09-21)
10 00
Windows MDM CSP policy expansion new 77

Adds six new Windows MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies. CSP-basedCSP-basedContent Security Policy, a W3C browser standard that restricts which scripts, styles, and resources a page may load, letting cyber tools enforce or audit site defenses against injection attacks. policies — Windows Wi-Fi Configuration Policy, Windows VPN Policy, Install Certificate Policy, Bluetooth Configuration Policy, Control Policy Conflict, and Windows Notification Service — plus a Create a Windows SCEP Profiles Policy for Simple Certificate Enrollment Protocol, and a generic Windows MDM CSPCSPContent Security Policy: a browser security mechanism, standardized by the W3C, that lets a web server declare which sources of scripts, styles, and other resources a page may load, blocking injection attacks. framework letting admins configure any company policy via a Configuration Service Provider. Later adds three Windows Personalization policies (Taskbar Customization, Screen Saver Configuration, Desktop and Lock Screen Settings) and Windows 11 25H2 support.

— Names every policy but no exact config path to enable them Release Notes 2025 (2026-09-21)Release Notes 2024 (2026-09-21)
11 00
JumpCloud Protect push and enrollment controls improved 75

Adds Push Bombing Mitigation to JumpCloud Protect, blocking more than one MFA push per resource within a 60-second window (limit adjustable or disabled in MFA Configurations); enables MFA by default for all new admin accounts; adds lock-screen push notification approval and denial (long-press on iOS, expand on Android) when biometric User Verification isn't required; and adds User Group Exclusion from Unified MFA Push enforcement, letting excluded groups use TOTP only.

Enable the Push Bombing Mitigation limit or disable it entirely for high-volume MFA environments where legitimate users may trigger multiple pushes.
📍In Admin Portal, go to Security › MFA Configurations, then adjust Maximum concurrent push attempts or toggle Push Bombing Mitigation off.
— Names the exact push limit window and a config example Release Notes 2025 (2026-09-21)Release Notes 2023 (2026-09-21)
12 00
Dynamic Device and User Groups new 75

Adds Dynamic Device Groups and Dynamic User Groups with attribute-based rules that automate group membership, with an optional review step before changes take effect. Adds Vendor and Public IP Address as new attributes for dynamic device group rules, and Manager as a new attribute for dynamic user group rules.

Set up a dynamic device group that automatically includes all devices with a known public IP range, removing manual membership updates.
📍In Admin Portal, go to Device Groups › New Group, enable Dynamic, and add a rule using the Public IP Address attribute with your target value.
— Names all three new attributes with a rule-building example Release Notes 2023 (2026-09-21)
13 05
Access Risk Detection (GA) new 75

Promotes Access Risk Detection to general availability, with behavior-based detection (unusual locations, impossible travel, unfamiliar devices/browsers, dormant accounts, abnormal login volumes), automatic MFA-based resolution, intelligent suppression for VPN/gateway logins, customizable risk factor thresholds, and email/Slack/webhook notifications.

— Enumerates detection signals and notification channels, no config steps Release Notes 2026 (2026-09-21)
14 00
Password Manager desktop app platform and region expansion improved 73

Adds a Debian (Ubuntu x86) build and later an RPM build for Linux (Red Hat/Fedora x64) with automatic updates, a native ARMARMA family of RISC processor architectures designed by Arm Holdings and licensed to chip makers worldwide; tools target it to run natively on mobile, embedded, and increasingly server and desktop hardware. build for macOS M-series (M1+) devices, EU region support selectable on the activation page and browser-extension pairing page, and installation/update via a new command template v.2.0.0 for Windows and macOS that removes the need for ongoing admin permissions after the initial run.

Onboard EU-region users so their vault data stays within the EU — select the region before activating to ensure data residency compliance.
📍On the Password Manager activation page, select EU as your region before entering your activation code, or on the browser extension pairing page, choose the EU region from the region selector.
— Names builds and template version with a region-selection example What’s New: Password Manager Desktop App (20…Release Notes 2023 (2026-09-21)
15 00
IPA and iOS/iPadOS BYOD enrollment enhancements new 73

Supports BYODBYODBring Your Own Device: a policy allowing employees to use personal phones, laptops, or tablets for work, creating a network of unmanaged endpoints that cyber tools must monitor or enforce policy on. enrollments of iOS/iPadOS 18+ devices via Account-Driven User Enrollment (ADUE), now generally available; adds two policies to control Apple Intelligence features on iOS/iPadOS 18.1 and macOS 15.1 Sequoia; adds Apple Business Manager as a new service discovery method for Account-driven Enrollment; and adds support for uploading and deploying IPA-based app packages to iOS and iPadOS devices via the JumpCloud Private Repository.

— Names each Apple MDM capability, no step-by-step enrollment guide Release Notes 2025 (2026-09-21)Release Notes 2024 (2026-09-21)
16 00
Commands feature: install trigger and running status new 73

Adds a Command After Agent Install command type so admins can specify enrollment commands to run immediately after the JumpCloud agent installs on new devices, and a Running column in the commands queue to distinguish commands waiting versus actively running.

Run a bootstrap script on every new device the moment the JumpCloud agent is installed, without waiting for the device to fully enroll.
📍In the Admin Portal, go to Commands › New Command › select command type Command After Agent Install, then write your onboarding script and save.
— Named command type with a full setup walkthrough Release Notes 2022 (2026-09-21)
17 00
Remote Assist remote desktop support new 72

Introduces JumpCloud Remote Assist for remote desktop access and support on Windows and macOS, available to all customers on JumpCloud Agent v1.61.0+ with no limits on session time or count. Adds support for Windows Server initial login screens (entering credentials and interacting with the login window), a Ctrl+Alt+Del shortcut from the Session Options toolbar, and user-switching between accounts during sessions on Windows. Later ships as two architecture-specific packages (ARM64ARM64A 64-bit CPU architecture from ARM Holdings, used in mobile chips, Apple Silicon, and cloud servers; cyber tools target it to run natively on those platforms instead of relying on emulation. for Apple Silicon, x86_64 for Intel Macs) and automatically detects and terminates existing unresponsive agent instances before establishing a new connection.

— Full before/after across releases but no admin UI path given Release Notes 2024 (2026-09-21)Release Notes 2023 (2026-09-21)Release Notes 2022 (2026-09-21)
18 00
Private Repository app version updates improved 72

Enables updating an existing application's binary in JumpCloud's Private Repository via Update Version on the application's Details tab, deploying to already-associated devices without re-creating the application or reassigning device groups.

— Names the exact UI control and its no-recreate behavior Release Notes 2026 (2026-09-21)
19 00
Knowledge Base Uploads for AI Assistant new 71

Adds Knowledge Base Uploads for AI Assistant, supporting PDF, TXT, DOCX, and MD files (10 MB max per file, 25 files max per org) in a private org-scoped knowledge base, with source attribution in chat responses.

— Exact file types and limits given, no upload steps Release Notes 2026 (2026-09-21)
20 00
Device management policies: macOS and Linux new 70

Adds a Managed Login Items policy for macOS 13 Ventura to prevent end users from disabling persistent software like the JumpCloud Agent, new software management policies to delay or block macOS Ventura upgrades on managed devices, and a Linux Forbidden Services policy to remove support for service types not required for everyday fleet operation.

Block unintended macOS Ventura upgrades across a managed fleet until your org is ready to migrate.
📍In the Admin Portal, go to Device Management › Policies › New Policy › macOS › select Delay or Block macOS Ventura, configure the deferral period, and bind it to the target device group.
— Names each policy with a worked deferral configuration example Release Notes 2022 (2026-09-21)
21 00
Multi-Tenant Portal and MSP management new 70

Adds MSPMSPA managed service provider, a company that operates IT services for customers. It may manage systems, networks, backups or other ongoing technology needs. Create Custom Command Templates so admins can build reusable command templates applied across tenant organizations; ability to run ad hoc commands on devices directly from the Multi-Tenant Portal and Admin Portal; user and device management in the Multi-Tenant Portal including creating users/devices, exporting lists, and bulk actions on users; an MTP Homepage widget showing aggregate reports across all tenant orgs from the last 24 hours with bulk JSON or CSV export; and a Global MFA Requirement setting under Account > Settings in the Multi-Tenant Portal for bulk MFA enforcement across all MSP admins.

— Names each MTP surface and export formats, no exact click path Release Notes 2025 (2026-09-21)Release Notes 2023 (2026-09-21)Release Notes 2022 (2026-09-21)
22 00
Asset Management field and cloning enhancements improved 68

Adds a File Upload custom field type and an Invoice file-upload field to Asset Management asset records, a Disk Size field under Hardware Specifications and Local Device Accounts under Device Management on the Asset Detail page, and asset cloning to streamline creating assets with shared hardware configurations, automatically clearing unique fields to prevent duplicates.

— Names each new field and the cloning behavior Release Notes 2025 (2026-09-21)
23 00
SCIM group management mapping UI new 68

Expands SCIMSCIMA standard for creating, updating and removing user accounts and groups across applications. It helps an identity provider keep access in sync when people join, change roles or leave. group management with a dedicated mapping UI supporting name overrides, extended schema mapping, side-by-side object transformation previews, flexible custom validation with selectable test users and groups, and automatic sync logging to Directory Insights.

— Enumerates mapping capabilities, no field-level example Release Notes 2026 (2026-09-21)
24 00
Device Monitoring and Alerting new 67

Introduces JumpCloud Device Monitoring and Alerting, enabling customizable alerts for device health (availability, software, uptime, disk, battery), policy/command/software management results, user addition and privilege elevation directory events, and command execution monitoring across Windows, macOS, and Linux endpoints.

— Enumerates monitored signals but no alert-setup steps Release Notes 2025 (2026-09-21)
25 00
Admin role management: Asset Manager and Custom Admin Roles new 67

Adds an Asset Manager built-in role in the Admin Portal giving dedicated access to Asset Management without broader Admin Portal access, and Custom Admin Roles for direct organizations enabling granular role configuration based on specific responsibilities.

Grant a procurement team member access to Asset Management without exposing other Admin Portal capabilities.
📍In the Admin Portal, go to Administrators › Roles, assign the built-in Asset Manager role to the user, then confirm the user can access Asset Management but not other admin areas.
— Names the built-in role with a walkthrough assigning it Release Notes 2025 (2026-09-21)
26 00
RADIUS and EAP authentication enhancements improved 67

Supports certificate (passwordless) based authentication for JumpCloud RADIUSRADIUSA networking protocol for centralised authentication, authorisation, and accounting of users connecting to a network, maintained under IETF standards. Cyber tools use it to integrate with existing enterprise identity infrastructure. accepting certificates from any X.509X.509A standard defining the format of public-key certificates, maintained by the ITU-T, used by cyber tools to authenticate identities and establish encrypted connections over TLS and similar protocols.-compliant certificate authority; adds RADIUS authentication using Azure AD as the identity provideridentity providerA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems.; supports multiple IP addresses on a single RADIUS server configuration; and adds faster reconnect for EAP connections by letting recently authenticated devices skip the full security handshake using a secure cached session credential.

— Names auth mechanisms across releases, no CLI setup given Release Notes 2026 (2026-09-21)Release Notes 2025 (2026-09-21)Release Notes 2022 (2026-09-21)
27 00
Shadow AI Dashboard new 66

Adds a Shadow AI Dashboard — a dedicated UI for managing generative AI app usage — with insights into AI adoption, usage by department, and most active users, under the renamed AI & SaaS ManagementSaaS ManagementA category of software that tracks, governs, and optimises an organisation's portfolio of SaaS subscriptions, covering spend, access, and licence utilisation from a single control plane. product.

Detect unauthorized or unmanaged generative AI app usage across your org before it becomes a shadow IT risk.
📍In the Admin Portal, go to AI & SaaS Management › Shadow AI Dashboard to review AI adoption by department, set approval or restriction policies on discovered AI apps, and identify most active users.
— Named UI with a navigation example, limited mechanism detail Release Notes 2025 (2026-09-21)
28 00
API filter escaping for pipe characters improved 65

Adds support for escaping the pipe | character with a backslash \ inside in and nin API filter search terms, e.g. GET /systemusers?filter=department:$in:....

— Names exact endpoint and escape syntax, no broader use case Release Notes 2024 (2026-09-21)
29 06
Android device management: zero-touch, EMM, and Android 15 new 65

Adds Android zero-touch enrollment for fully managed, dedicated, and company-owned devices with a work profile; Android EMMEMMEnterprise Mobility Management: a category of software for enforcing policies on mobile devices, apps, and data, giving cyber tools a hook to monitor or control endpoints without managing the full OS. support for fully managed and dedicated devices with 7 new policies including Factory Reset Protection, Kiosk Mode, and System Updates; and full Android 15 MDM support.

— Names enrollment modes and example policies, no config path Release Notes 2024 (2026-09-21)Release Notes 2023 (2026-09-21)
30 00
SaaS Management app discovery and governance new 64

Introduces SaaS Management: discover and track SaaS app usage via the JumpCloud Go browser extension, categorize apps, restrict access with warning or blocking messaging, and manually track licensing. Later enables app discovery without requiring JumpCloud Go to be enabled as a feature (the browser extension is still required), and lets admins manually add applications for tracking, warning, and blocking before automatic discovery occurs.

— Explains discovery mechanism but no navigation path Release Notes 2024 (2026-09-21)
31 00
RSS feeds for release and bug fix announcements new 64

Adds RSS feed subscriptions for JumpCloud Help Center content, with dedicated feeds for product release notes, agent release notes, and bug fix reports via Subscribe to Support RSS Feed, Subscribe to Release Notes and Bug Fixes RSS Feed, Subscribe to Agent Release Notes RSS Feed, and Subscribe to Bug Fixes RSS Feed.

Subscribe to only the product and agent release notes RSS feed to track new JumpCloud capabilities in your RSS reader.
📍1. Go to the JumpCloud Release Notes and Bug Fixes page. 2. Click Subscribe to Release Notes and Bug Fixes RSS Feed. 3. Open your RSS reader and add the returned feed URL to begin receiving product release notes and agent release notes automatically.
— Names each feed with a subscription walkthrough Subscribe to New Feature and Product Update …
32 00
Windows App Catalog new 62

Introduces the JumpCloud App Catalog for Windows with 10 apps — 1Password, Adobe Acrobat Reader DC, Chrome, Firefox, Google Drive, OneDrive, OpenVPN, Slack, Webex, and Zoom — with auto-update support.

— Lists all ten apps but no enablement steps Release Notes 2024 (2026-09-21)
33 00
Browser and OS patch management reporting new 60

Adds a universal Google Chrome Browser Patch Management policy covering macOS, Windows, and Linux from a single policy, an OS Patch Management Policy Report showing which policies are applied to each device and their compliance status, and a Browser Patch Management Policy Report showing devices' current browser versions for compliance audits.

— Names the cross-platform policy and both reports, no export detail Release Notes 2022 (2026-09-21)
thinner coverage below
34 00
Access Requests new 59

Adds JumpCloud Access Requests with configurable approval flows, request management in the Admin Portal, tracking via Directory Insights, and webhook integration with ticketing systems such as JIRA.

— Names the ticketing integration but no request-creation steps Release Notes 2025 (2026-09-21)
35 07
Conditional Access disk encryption condition new 59

Adds a new device condition in Conditional AccessConditional AccessA policy enforcement framework, most commonly Microsoft's in Entra ID, that grants or blocks resource access based on signals like user identity, device state, and location — letting cyber tools trigger step-up auth or deny risky sessions. Policies allowing or denying access based on whether disk encryption is enabled or disabled (FileVaultFileVaultApple's full-disk encryption feature built into macOS, using XTS-AES-128 to encrypt the startup volume so that data is unreadable without the user's credentials. on macOS, system drive on Windows, root disk on Linux).

— Names the platform-specific encryption checks, no policy-build steps Release Notes 2022 (2026-09-21)
36 00
MDM certificate and token expiration visibility new 58

Adds the MDM Upcoming Certificate Expirations homepage widget for at-a-glance lifecycle visibility into MDM, Apple VPP, and Automated Device Enrollment certificates, and an MDM Certificate & Token Status Board in the Admin Portal and Multi-Tenant Portal providing expiry status for MDM, SSO, and AzureAD tokens and certificates with an aggregate MSP view.

— Names both surfaces tracked, no remediation steps Release Notes 2023 (2026-09-21)Release Notes 2022 (2026-09-21)
37 00
Admin Portal quick search shortcut new 55

Adds a Cmd/Ctrl + K keyboard shortcut in the Admin Portal for fast search and navigation between areas.

— Exact shortcut given, minimal further detail needed Release Notes 2024 (2026-09-21)
38 00
SentinelOne agent deployment via Commands new 54

Adds SentinelOne integration via the Commands feature, enabling download and installation of the SentinelOne Agent on macOS, Windows, and Linux devices.

— Names the vendor and platforms, no command syntax given Release Notes 2022 (2026-09-21)
39 00
Directory Insights filter granularity improved 54

Expands Directory Insights Application Name and Resource Name filters into sub-categories: Application Display Name, Application Display Label, Resource Display Name, and Resource Display Label.

— Names all four new sub-filters, no query example Release Notes 2025 (2026-09-21)
40 00
AI Gateway third-party integrations new 53

Adds AI Gateway integration support for Miro, Mixpanel, Netlify, Supabase, Vercel, and Zoho, plus new user-facing AI Gateway connectors for Antigravity and Claude Code.

— Names all eight integrations, no setup mechanism given product docs
41 00
Schedule macOS software updates per device new 53

Adds ability to schedule macOS software updates for a single device from the Admin Portal, including configurable install timeframe and user deferral options.

— Describes scheduling options without exact navigation path Release Notes 2022 (2026-09-21)
42 00
Background Tools remote troubleshooting new 52

Adds Background Tools remote command line and remote file transfer capabilities, letting admins access and troubleshoot end-user devices without interrupting active sessions.

— Names the capability without exact invocation steps Release Notes 2023 (2026-09-21)
43 00
New MFA methods: TOTP for LDAP and Duo Universal Prompt new 50

Adds Time-based One-time Password (TOTP) as a supported MFA factor for LDAP applications alongside JumpCloud Protect push notifications, and adds Duo Universal Prompt support for JumpCloud MFA ahead of Duo's March 30, 2024 deprecation of the traditional prompt.

— Names both methods and Duo's deprecation date, no setup steps Release Notes 2023 (2026-09-21)Release Notes 2022 (2026-09-21)
44 08
JumpCloud MCP Server (GA) new 50

Adds JumpCloud MCP Server (GA) to connect compatible AI tools to JumpCloud using the Model Context ProtocolModel Context ProtocolModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface., enabling interaction with org data and admin capabilities from AI tools and code editors.

— Describes purpose but no connection steps or tool list Release Notes 2025 (2026-09-21)
45 00
Self-Service Account Provisioning new 50

Adds Self-Service Account Provisioning allowing new users to bind their account to JumpCloud-managed macOS and Windows devices from the login window, including wireless connectivity controls.

— Describes the login-window flow without setup steps Release Notes 2023 (2026-09-21)
46 00
Support case portal improved 50

Adds the JumpCloud Case Portal to centralize support cases and ideas from all administrators in one filterable, searchable view; later renamed/expanded into a Customer Portal with sortable columns for Last Modified Date, Account, and Requester, plus attachment uploads and inline ticket editing.

— Names sortable columns, no ticket-creation steps Release Notes 2025 (2026-09-21)Release Notes 2022 (2026-09-21)
47 09
Password Manager launch: encrypted sync, autofill, and sharing new 48

Launches JumpCloud Password Manager with end-to-end encrypted local sync, a browser extension for autofill, secure password sharing, and password generation — no master password required.

— Describes launch features but no config keys or exact steps Release Notes 2022 (2026-09-21)
48 00
Password policy and self-service reset behavior improved 47

Changes the default password policy compliance behavior so users are forced to comply on their next password change rather than immediately, and enables admins to allow users to reset passwords from the User Portal when MFA is enforced but not yet enrolled.

— Describes behavior change without the config toggle name Release Notes 2025 (2026-09-21)
49 00
AI Commands Builder (GA) new 45

Adds AI Commands Builder (GA) to create and refine commands for Windows, macOS, and Linux devices using natural-language prompts within the Admin Portal.

— States capability without a worked prompt example Release Notes 2025 (2026-09-21)
50 00
Insights Rules frequency-based alert triggers new 45

Adds alert trigger conditions based on event frequency to Insights Rules, allowing alerts to fire only when defined thresholds are met rather than on every individual event.

— Describes threshold behavior without exact configuration UI Release Notes 2026 (2026-09-21)
51 00
User Portal Applications tabs improved 43

Adds All Applications and Favorites tabs to the Applications area of the User Portal, with starring and responsive search on both tabs.

— Names the tabs and search behavior, no navigation path Release Notes 2023 (2026-09-21)
52 00
Password Manager .appimage build discontinued deprecated 40

The .appimage Linux build is no longer supported as of April 18, 2024; users must migrate to the RPM or Debian builds.

— Clear cutover date but no migration command What’s New: Password Manager Desktop App (20…
53 00
Password Date column on Users page improved 40

Adds Password Date (last password set date) as a viewable column on the Users page in Admin Portal.

— Names the exact column with clear location, thin scope Release Notes 2023 (2026-09-21)
54 10
SaaS License Management new 38

Adds SaaS License Management (GA) for automated license tracking, usage optimization, and cost management via SaaS application connectors.

— Names the capability without connector list or setup steps Release Notes 2025 (2026-09-21)
55 00
Service Account for APIs new 35

Adds Service Account for APIs, enabling API service accounts that are not tied to individual Admin accounts.

— No creation steps or scoping detail given Release Notes 2025 (2026-09-21)
56 00
ADI support for Windows Server 2022 new 35

Adds ADI support for Windows Server 2022 domain controllers and member servers without workarounds or reduced functionality.

— States support without configuration guidance Release Notes 2023 (2026-09-21)
57 00
Identity First Login flow new 33

Introduces an Identity First Login flow for the User Portal and SSO applications, splitting login into a username screen followed by a separate password screen.

— Describes UX change without rollout or config detail Release Notes 2022 (2026-09-21)
58 00
OIDC single sign-on support new 32

Supports OpenID ConnectOpenID ConnectOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. (OIDC) Single Sign-On so users can authenticate to OIDC-capable web applications with a single set of credentials.

— States capability without configuration detail Release Notes 2022 (2026-09-21)
59 00
WebAuthn enrollment split into credential types improved 32

Splits WebAuthnWebAuthnA web standard that lets a browser prove possession of a passkey or hardware key to a website. It is the browser-facing part of FIDO2 sign-in. enrollment into separate credential types — Security Keys and Device authenticators — for clearer enrollment.

— No mention of where enrollment lives or migration effect Release Notes 2023 (2026-09-21)
60 00
M365/Azure AD security groups in directory integration improved 29

Adds M365/Azure AD security groups support to the JumpCloud Microsoft 365 directory integration.

— Bare capability statement, no sync scope detail Release Notes 2022 (2026-09-21)
61 00
Legacy Password Manager CSV export for Vault migration improved 25

Enhances CSV export for the legacy Password Manager with new data fields to support migration into JumpCloud Vault.

— No field names or migration steps given What’s New: Password Manager Desktop App (20…
62 00
Deel SSO Connector new 23

Adds Deel SSO Connector integration with JumpCloud.

— Bare integration name with no configuration detail Release Notes 2023 (2026-09-21)
63 00
Trials recap summary new 21

Adds JumpCloud Trials recap summarizing feature usage at the end of a trial.

— Bare description with no detail on delivery or contents Release Notes 2022 (2026-09-21)
64 00
Linux Fedora 37 agent support new 21

Adds JumpCloud agent support for Linux Fedora 37.

— Bare platform-support statement Release Notes 2022 (2026-09-21)
└──▷ ALSO FROM THESE RELEASES
└──▷ BREAKING ON UPGRADE
  • !
    The GET api/systemusers/reactivate/{id} endpoint is deprecated and now returns a 404 error; all callers must switch to POST https://console.jumpcloud.com/systemusers/reactivate/{id} with activation_email in the request body.
  • !
    Dynamic Groups memberQuery filter operators $le and $ge are deprecated in favor of $lte and $gte; existing API scripts using $le or $ge must be updated (Admin Portal-managed Dynamic Groups are not affected).
  • !
    JumpCloud LDAP no longer accepts connections using CBC-mode or SHA-1/SHA-256 integrity cipher suites (including RSARSAThe oldest widely used public-key algorithm, for both encryption and signatures, with security resting on factoring large numbers. A large quantum computer would break it. and ECDHE key exchange variants); only the five explicitly listed GCM cipher suites are permitted. Any LDAP client configured to use only the removed suites will fail to connect.
  • !
    The .appimage Linux build is no longer supported as of April 18, 2024; users must migrate to the RPM or Debian builds.
Release Notes 2022 (2026-09-21) Blog Stable

JumpCloud 2022 adds certificate-based RADIUSRADIUSA networking protocol for centralised authentication, authorisation, and accounting of users connecting to a network, maintained under IETF standards. Cyber tools use it to integrate with existing enterprise identity infrastructure. auth, TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip. for LDAPLDAPA protocol for querying and modifying directory services, standardized under RFC 4511, commonly used to look up users, groups, and attributes in corporate identity stores like Active Directory. MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone., Remote Assist, Password Manager, and a new Command After Agent Install type.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Run a bootstrap script on every new device the moment the JumpCloud agent is installed, without waiting for the device to fully enroll.
📍In the Admin Portal, go to Commands › New Command › select command type Command After Agent Install, then write your onboarding script and save.
Block unintended macOS Ventura upgrades across a managed fleet until your org is ready to migrate.
📍In the Admin Portal, go to Device Management › Policies › New Policy › macOS › select Delay or Block macOS Ventura, configure the deferral period, and bind it to the target device group.
  • Adds Command After Agent Install command type so admins can specify enrollment commands to run immediately after the JumpCloud agent is installed on new devices.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds a Running column to the commands queue to distinguish commands waiting in the queue from those actively running.
  • Supports certificate (passwordless) based authentication for JumpCloud RADIUS, accepting certificates from any X.509X.509A standard defining the format of public-key certificates, maintained by the ITU-T, used by cyber tools to authenticate identities and establish encrypted connections over TLS and similar protocols.-compliant certificate authority.
  • Adds Time-based One-time Password (TOTP) as a supported MFA factor for LDAP applications, in addition to JumpCloud Protect push notifications.
  • Introduces JumpCloud Remote Assist for remote desktop access and support on Windows and macOS devices, available to all customers running JumpCloud Agent v1.61.0 or later, with no limits on time or number of sessions.
+20 more show less
  • Launches JumpCloud Password Manager with end-to-end encrypted local sync, browser extension auto-fill, secure password sharing, and password generation — no master password required.
  • Adds RADIUS authentication using Azure AD as the identity provideridentity providerA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems., enabling secure RADIUS access without managing users and passwords outside of Azure AD.
  • Adds a Linux Forbidden Services policy to remove support for service types not required for everyday Linux fleet operation.
  • Adds a Managed Login Items policy for macOS 13 Ventura to prevent end users from disabling persistent software such as the JumpCloud Agent.
  • Adds new software management policies to delay or block macOS Ventura upgrades on managed devices.
  • Adds a new device condition in Conditional AccessConditional AccessA policy enforcement framework, most commonly Microsoft's in Entra ID, that grants or blocks resource access based on signals like user identity, device state, and location — letting cyber tools trigger step-up auth or deny risky sessions. Policies allowing or denying access based on whether disk encryption (FileVaultFileVaultApple's full-disk encryption feature built into macOS, using XTS-AES-128 to encrypt the startup volume so that data is unreadable without the user's credentials. on macOS, system drive on Windows, root disk on Linux) is enabled or disabled.
  • Supports OpenID ConnectOpenID ConnectOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. (OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user.) Single Sign-On so users can authenticate to OIDC-capable web applications with a single set of credentials.
  • Adds JumpCloud agent support for Linux Fedora 37.
  • Adds MSPMSPA managed service provider, a company that operates IT services for customers. It may manage systems, networks, backups or other ongoing technology needs. Create Custom Command Templates so Managed Service Provider admins can create reusable command templates and apply them across tenant organizations.
  • Adds a universal Google Chrome Browser Patch Management policy covering macOS, Windows, and Linux devices from a single policy.
  • Introduces an Identity First Login flow for the User Portal and SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. applications, splitting login into a username screen followed by a password screen.
  • Adds SentinelOne integration via the Commands feature, enabling download and installation of the SentinelOne Agent on macOS, Windows, and Linux devices.
  • Adds the OS Patch Management Policy Report showing which policies are applied to each device and their current compliance status.
  • Adds the Browser Patch Management Policy Report for evidence during compliance audits, showing devices' current browser versions.
  • Adds the MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies. Upcoming Certificate Expirations homepage widget providing at-a-glance lifecycle visibility into MDM, Apple VPP, and Automated Device Enrollment certificates.
  • Adds MTP Homepage widget showing aggregate reports available to download across all tenant organizations in the last 24 hours, with bulk JSON or CSV export.
  • Adds ability to schedule macOS software updates for a single device from the Admin Portal, including configurable install timeframe and user deferral options.
  • Adds the JumpCloud Case Portal to centralize support cases and ideas from all administrators in one filterable, searchable view.
  • Adds JumpCloud Trials recap summarizing feature usage at the end of a trial.
  • Adds M365/Azure AD security groups support to the JumpCloud Microsoft 365 directory integration.
  • Adds Command After Agent Install command type so admins can specify enrollment commands to run immediately after the JumpCloud agent is installed on new devices.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Supports certificate (passwordless) based authentication for JumpCloud RADIUS, accepting certificates from any X.509X.509A standard defining the format of public-key certificates, maintained by the ITU-T, used by cyber tools to authenticate identities and establish encrypted connections over TLS and similar protocols.-compliant certificate authority.
  • Adds Time-based One-time Password (TOTP) as a supported MFA factor for LDAP applications, in addition to JumpCloud Protect push notifications.
  • Introduces JumpCloud Remote Assist for remote desktop access and support on Windows and macOS devices, available to all customers running JumpCloud Agent v1.61.0 or later, with no limits on time or number of sessions.
  • Launches JumpCloud Password Manager with end-to-end encrypted local sync, browser extension auto-fill, secure password sharing, and password generation — no master password required.
+15 more show less
  • Adds RADIUS authentication using Azure AD as the identity provideridentity providerA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems., enabling secure RADIUS access without managing users and passwords outside of Azure AD.
  • Adds a Linux Forbidden Services policy to remove support for service types not required for everyday Linux fleet operation.
  • Adds a Managed Login Items policy for macOS 13 Ventura to prevent end users from disabling persistent software such as the JumpCloud Agent.
  • Adds new software management policies to delay or block macOS Ventura upgrades on managed devices.
  • Supports OpenID ConnectOpenID ConnectOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. (OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user.) Single Sign-On so users can authenticate to OIDC-capable web applications with a single set of credentials.
  • Adds MSPMSPA managed service provider, a company that operates IT services for customers. It may manage systems, networks, backups or other ongoing technology needs. Create Custom Command Templates so Managed Service Provider admins can create reusable command templates and apply them across tenant organizations.
  • Adds a universal Google Chrome Browser Patch Management policy covering macOS, Windows, and Linux devices from a single policy.
  • Introduces an Identity First Login flow for the User Portal and SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. applications, splitting login into a username screen followed by a password screen.
  • Adds SentinelOne integration via the Commands feature, enabling download and installation of the SentinelOne Agent on macOS, Windows, and Linux devices.
  • Adds the OS Patch Management Policy Report showing which policies are applied to each device and their current compliance status.
  • Adds the Browser Patch Management Policy Report for evidence during compliance audits, showing devices' current browser versions.
  • Adds the MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies. Upcoming Certificate Expirations homepage widget providing at-a-glance lifecycle visibility into MDM, Apple VPP, and Automated Device Enrollment certificates.
  • Adds MTP Homepage widget showing aggregate reports available to download across all tenant organizations in the last 24 hours, with bulk JSON or CSV export.
  • Adds the JumpCloud Case Portal to centralize support cases and ideas from all administrators in one filterable, searchable view.
  • Adds JumpCloud Trials recap summarizing feature usage at the end of a trial.

5 more in Everything

Release Notes 2023 (2026-09-21) Blog Stable

JumpCloud 2023 adds passwordless Go auth, Android zero-touch enrollment, remote Background Tools, dynamic groups, and Duo Universal Prompt support.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Enable the Push Bombing Mitigation limit or disable it entirely for high-volume MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. environments where legitimate users may trigger multiple pushes.
📍In Admin Portal, go to Security › MFA Configurations, then adjust Maximum concurrent push attempts or toggle Push Bombing Mitigation off.
Set up a dynamic device group that automatically includes all devices with a known public IP range, removing manual membership updates.
📍In Admin Portal, go to Device Groups › New Group, enable Dynamic, and add a rule using the Public IP Address attribute with your target value.
  • Adds JumpCloud Go browser extension enabling hardware-protected, phishing-resistant passwordless login on managed macOS and Windows devices using Apple Touch ID and Windows HelloWindows HelloA Microsoft authentication framework built into Windows 10 and later that lets users sign in via biometrics or PIN instead of passwords, giving cyber tools a phishing-resistant credential layer to build on..⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds JumpCloud Go as the default MFA method for conditional access policies on SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. apps, with user and device verification on every application session.
  • Adds AWS Verified Access integration supporting JumpCloud Go, SAMLSAMLA standard for passing signed identity assertions between a sign-in provider and an application. It lets a user sign in through their organization's identity provider., and OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. SSO as a VPNVPNA virtual private network, an encrypted tunnel between devices or networks. It carries traffic across another network as though the endpoints had a private connection.-less remote access solution for corporate web resources.
  • Adds Android zero-touch enrollment support for fully managed, dedicated, and company-owned Android devices with a work profile.
  • Adds Background Tools remote command line and remote file transfer capabilities, letting admins access and troubleshoot end-user devices without interrupting active sessions.
+23 more show less
  • Adds Self-Service Account Provisioning allowing new users to bind their account to JumpCloud-managed macOS and Windows devices from the login window, including wireless connectivity controls.
  • Adds MDM Certificate & Token Status Board in the Admin Portal and Multi-Tenant Portal, providing expiry status for MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies., SSO, and AzureAD tokens and certificates with an aggregate MSPMSPA managed service provider, a company that operates IT services for customers. It may manage systems, networks, backups or other ongoing technology needs. view.
  • Adds Dynamic Device Groups and Dynamic User Groups with attribute-based rules automating group membership, with an optional review step before changes take effect.
  • Adds Vendor and Public IP Address as new attributes for dynamic device group rules.
  • Adds Manager as a new attribute for dynamic user group rules.
  • Adds Password Manager Shared Folders management in the admin portal, with per-folder manual sync and improved folder creation workflow in the desktop application.
  • Adds JumpCloud Password Manager Cloud Backups giving admins a secure fallback to minimize vault data loss.
  • Adds Push Bombing Mitigation to JumpCloud Protect, blocking more than one MFA push per resource within a sixty-second window; admins can adjust the limit or disable the restriction in MFA Configurations.
  • Adds Duo Universal Prompt support for JumpCloud MFA, ahead of Duo's March 30, 2024 deprecation of the traditional prompt.
  • Adds ability for admins to run ad hoc commands on devices directly from the Multi-Tenant Portal and Admin Portal.
  • Adds user and device management to the Multi-Tenant Portal, including creating new users and devices, exporting user and device lists, and bulk actions on users.
  • Adds ADI support for Windows Server 2022 domain controllers and member servers without workarounds or reduced functionality.
  • Adds Deel SSO Connector integration with JumpCloud.
  • Adds Password Manager Desktop App Debian (Ubuntu x86) build for Linux with automatic update support.
  • Adds support for Remote Assist on Windows Server initial login screens, allowing admins to enter credentials and interact with the login window.
  • Adds Ctrl+Alt+Del shortcut from the Session Options toolbar during Remote Assist sessions on Windows devices.
  • Adds user-switching capability between accounts during Remote Assist sessions on Windows devices.
  • Adds Android EMMEMMEnterprise Mobility Management: a category of software for enforcing policies on mobile devices, apps, and data, giving cyber tools a hook to monitor or control endpoints without managing the full OS. support for fully managed and dedicated devices with 7 new policies including Factory Reset Protection, Kiosk Mode, and System Updates.
  • Adds Password Date (last password set date) as a viewable column on the Users page in Admin Portal.
  • Splits WebAuthnWebAuthnA web standard that lets a browser prove possession of a passkey or hardware key to a website. It is the browser-facing part of FIDO2 sign-in. enrollment into separate credential types — Security Keys and Device authenticators — for clearer enrollment.
  • Adds All Applications and Favorites tabs to the Applications area of the User Portal, with starring and responsive search on both tabs.
  • Enables MFA by default for all new admin accounts logging into the Admin Portal.
  • Adds lock-screen push notification approval and denial in JumpCloud Protect (long-press on iOS, expand on Android) when User Verification biometrics is not required.
  • Adds JumpCloud Go browser extension enabling hardware-protected, phishing-resistant passwordless login on managed macOS and Windows devices using Apple Touch ID and Windows HelloWindows HelloA Microsoft authentication framework built into Windows 10 and later that lets users sign in via biometrics or PIN instead of passwords, giving cyber tools a phishing-resistant credential layer to build on..⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds AWS Verified Access integration supporting JumpCloud Go, SAMLSAMLA standard for passing signed identity assertions between a sign-in provider and an application. It lets a user sign in through their organization's identity provider., and OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. SSO as a VPNVPNA virtual private network, an encrypted tunnel between devices or networks. It carries traffic across another network as though the endpoints had a private connection.-less remote access solution for corporate web resources.
  • Adds Android zero-touch enrollment support for fully managed, dedicated, and company-owned Android devices with a work profile.
  • Adds Background Tools remote command line and remote file transfer capabilities, letting admins access and troubleshoot end-user devices without interrupting active sessions.
  • Adds Self-Service Account Provisioning allowing new users to bind their account to JumpCloud-managed macOS and Windows devices from the login window, including wireless connectivity controls.
+11 more show less
  • Adds MDM Certificate & Token Status Board in the Admin Portal and Multi-Tenant Portal, providing expiry status for MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies., SSO, and AzureAD tokens and certificates with an aggregate MSPMSPA managed service provider, a company that operates IT services for customers. It may manage systems, networks, backups or other ongoing technology needs. view.
  • Adds Dynamic Device Groups and Dynamic User Groups with attribute-based rules automating group membership, with an optional review step before changes take effect.
  • Adds Password Manager Shared Folders management in the admin portal, with per-folder manual sync and improved folder creation workflow in the desktop application.
  • Adds JumpCloud Password Manager Cloud Backups giving admins a secure fallback to minimize vault data loss.
  • Adds Push Bombing Mitigation to JumpCloud Protect, blocking more than one MFA push per resource within a sixty-second window; admins can adjust the limit or disable the restriction in MFA Configurations.
  • Adds Duo Universal Prompt support for JumpCloud MFA, ahead of Duo's March 30, 2024 deprecation of the traditional prompt.
  • Adds ability for admins to run ad hoc commands on devices directly from the Multi-Tenant Portal and Admin Portal.
  • Adds Deel SSO Connector integration with JumpCloud.
  • Adds Ctrl+Alt+Del shortcut from the Session Options toolbar during Remote Assist sessions on Windows devices.
  • Adds user-switching capability between accounts during Remote Assist sessions on Windows devices.
  • Adds Password Date (last password set date) as a viewable column on the Users page in Admin Portal.

12 more in Everything

Release Notes 2024 (2026-09-21) Blog Stable

JumpCloud 2024 adds Device Monitoring & Alerting, Windows App Catalog, SaaS ManagementSaaS ManagementA category of software that tracks, governs, and optimises an organisation's portfolio of SaaS subscriptions, covering spend, access, and licence utilisation from a single control plane., BYODBYODBring Your Own Device: a policy allowing employees to use personal phones, laptops, or tablets for work, creating a network of unmanaged endpoints that cyber tools must monitor or enforce policy on. iOS 18 MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies., CSPCSPContent Security Policy: a browser security mechanism, standardized by the W3C, that lets a web server declare which sources of scripts, styles, and other resources a page may load, blocking injection attacks. policy framework, and more.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ TRY IT
Reactivate a user account after migrating scripts from the deprecated GET endpoint to the new POST endpoint.
$ POST https://console.jumpcloud.com/systemusers/reactivate/{id}
{
  "activation_email": "{email_address}"
}
  • Adds GET /systemusers?filter=department:$in:... support for escaping the pipe | character with backslash \ inside in and nin API filter search terms.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds six new Windows MDM CSP-basedCSP-basedContent Security Policy, a W3C browser standard that restricts which scripts, styles, and resources a page may load, letting cyber tools enforce or audit site defenses against injection attacks. policies: Windows Wi-Fi Configuration Policy, Windows VPNVPNA virtual private network, an encrypted tunnel between devices or networks. It carries traffic across another network as though the endpoints had a private connection. Policy, Install Certificate Policy, Bluetooth Configuration Policy, Control Policy Conflict, and Windows Notification Service.
  • Adds a new Windows MDM policy to configure Simple Certificate Enrollment Protocol (SCEPSCEPSimple Certificate Enrollment Protocol, an IETF-standardized protocol for automating X.509 certificate issuance and renewal over HTTP, letting cyber tools request PKI certificates from a CA without manual intervention.) for Windows devices via Create a Windows SCEP Profiles Policy.
  • Replaces the deprecated GET api/systemusers/reactivate/{id} endpoint with POST https://console.jumpcloud.com/systemusers/reactivate/{id} accepting activation_email in the request body.
  • Replaces deprecated $le and $ge Dynamic Groups memberQuery filter operators with $lte and $gte to align with the rest of the API list endpoints.
+18 more show less
  • Introduces JumpCloud Device Monitoring and Alerting, enabling customizable alerts for device health (availability, software, uptime, disk, battery), policy/command/software management results, user addition and privilege elevation directory events, and command execution monitoring across Windows, macOS, and Linux endpoints.
  • Introduces the JumpCloud App Catalog for Windows with 10 apps (1Password, Adobe Acrobat Reader DC, Chrome, Firefox, Google Drive, OneDrive, OpenVPN, Slack, Webex, Zoom) and auto-update support.
  • Introduces SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally. Management: discover and track SaaS app usage via the JumpCloud Go browser extension, categorize apps, restrict access with warning or blocking messaging, and manually track licensing.
  • Supports BYOD enrollments of iOS/iPadOS 18+ devices via Account-Driven User Enrollment (ADUE), now in general availability.
  • Adds two new JumpCloud MDM policies to control Apple Intelligence features on iOS/iPadOS 18.1 and macOS 15.1 Sequoia.
  • Adds full Android 15 MDM support.
  • Adds a new Windows MDM CSP framework allowing IT admins to configure any company policies using a Configuration Service Provider (CSP).
  • Adds support for multiple IP addresses on a single RADIUSRADIUSA networking protocol for centralised authentication, authorisation, and accounting of users connecting to a network, maintained under IETF standards. Cyber tools use it to integrate with existing enterprise identity infrastructure. server configuration.
  • Enables SaaS Management app discovery without requiring JumpCloud Go to be enabled as a feature (browser extension still required if JumpCloud Go is not enabled).
  • Allows admins to manually add applications to SaaS Management for tracking, warning, and blocking before automatic discovery occurs.
  • Adds a Cmd/Ctrl + K keyboard shortcut in the Admin Portal for fast search and navigation between areas.
  • Adds a User Portal > Security > Multi-Factor Authentication section for managing JumpCloud Go sessions on user devices.
  • Adds Global MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. Requirement setting under Account > Settings in the Multi-Tenant Portal, enabling bulk MFA enforcement for all MSPMSPA managed service provider, a company that operates IT services for customers. It may manage systems, networks, backups or other ongoing technology needs. admins.
  • Changes the default password policy compliance behavior so users are forced to comply on their next password change rather than immediately.
  • Enables admins to allow users to reset passwords from the User Portal when MFA is enforced but not yet enrolled.
  • JumpCloud Go sessions now automatically refresh when a user logs in to or unlocks their managed device.
  • Remote Assist now ships as two separate architecture-specific packages: ARM64ARM64A 64-bit CPU architecture from ARM Holdings, used in mobile chips, Apple Silicon, and cloud servers; cyber tools target it to run natively on those platforms instead of relying on emulation. for Apple Silicon and x86_64 for Intel Macs.
  • Remote Assist now automatically detects and terminates existing unresponsive agent instances on the end-user device before establishing a new connection.
  • Adds GET /systemusers?filter=department:$in:... support for escaping the pipe | character with backslash \ inside in and nin API filter search terms.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds six new Windows MDM CSP-basedCSP-basedContent Security Policy, a W3C browser standard that restricts which scripts, styles, and resources a page may load, letting cyber tools enforce or audit site defenses against injection attacks. policies: Windows Wi-Fi Configuration Policy, Windows VPNVPNA virtual private network, an encrypted tunnel between devices or networks. It carries traffic across another network as though the endpoints had a private connection. Policy, Install Certificate Policy, Bluetooth Configuration Policy, Control Policy Conflict, and Windows Notification Service.
  • Adds a new Windows MDM policy to configure Simple Certificate Enrollment Protocol (SCEPSCEPSimple Certificate Enrollment Protocol, an IETF-standardized protocol for automating X.509 certificate issuance and renewal over HTTP, letting cyber tools request PKI certificates from a CA without manual intervention.) for Windows devices via Create a Windows SCEP Profiles Policy.
  • Introduces JumpCloud Device Monitoring and Alerting, enabling customizable alerts for device health (availability, software, uptime, disk, battery), policy/command/software management results, user addition and privilege elevation directory events, and command execution monitoring across Windows, macOS, and Linux endpoints.
  • Introduces the JumpCloud App Catalog for Windows with 10 apps (1Password, Adobe Acrobat Reader DC, Chrome, Firefox, Google Drive, OneDrive, OpenVPN, Slack, Webex, Zoom) and auto-update support.
+7 more show less
  • Introduces SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally. Management: discover and track SaaS app usage via the JumpCloud Go browser extension, categorize apps, restrict access with warning or blocking messaging, and manually track licensing.
  • Supports BYOD enrollments of iOS/iPadOS 18+ devices via Account-Driven User Enrollment (ADUE), now in general availability.
  • Adds two new JumpCloud MDM policies to control Apple Intelligence features on iOS/iPadOS 18.1 and macOS 15.1 Sequoia.
  • Adds a new Windows MDM CSP framework allowing IT admins to configure any company policies using a Configuration Service Provider (CSP).
  • Adds a Cmd/Ctrl + K keyboard shortcut in the Admin Portal for fast search and navigation between areas.
  • Adds a User Portal > Security > Multi-Factor Authentication section for managing JumpCloud Go sessions on user devices.
  • Adds Global MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. Requirement setting under Account > Settings in the Multi-Tenant Portal, enabling bulk MFA enforcement for all MSPMSPA managed service provider, a company that operates IT services for customers. It may manage systems, networks, backups or other ongoing technology needs. admins.

11 more in Everything

└──▷ BREAKING ON UPGRADE
  • !The GET api/systemusers/reactivate/{id} endpoint is deprecated and now returns a 404 error; all callers must switch to POST https://console.jumpcloud.com/systemusers/reactivate/{id} with activation_email in the request body.
  • !Dynamic Groups memberQuery filter operators $le and $ge are deprecated in favor of $lte and $gte; existing API scripts using $le or $ge must be updated (Admin Portal-managed Dynamic Groups are not affected).
Release Notes 2025 (2026-09-21) Blog Stable

JumpCloud adds Access Requests, MCP ServerMCP ServerA process that exposes tools, resources, and prompts over Anthropic's Model Context Protocol, letting an AI agent call external APIs or local services through a standardised interface., Shadow AI Dashboard, forced re-auth/AMR claimsAMR claimsAssertions made by a CPU about its own current operating mode and privilege level, used by attestation systems to verify that code is running in a trusted execution environment., Custom Admin Roles, and IPA file deployment in 2025 releases.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Audit which authentication factors were used during an SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. login by inspecting the AMR claim in a JumpCloud-issued OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. ID token — useful for verifying Zero Trust policy compliance.
📍In the Admin Portal, go to Directory Insights and filter by event type sso_auth to view forced re-authentication events and confirm AMR tags (e.g., password + OTP) for each login.
Detect unauthorized or unmanaged generative AI app usage across your org before it becomes a shadow IT risk.
📍In the Admin Portal, go to AI & SaaS Management › Shadow AI Dashboard to review AI adoption by department, set approval or restriction policies on discovered AI apps, and identify most active users.
Grant a procurement team member access to Asset Management without exposing other Admin Portal capabilities.
📍In the Admin Portal, go to Administrators › Roles, assign the built-in Asset Manager role to the user, then confirm the user can access Asset Management but not other admin areas.
  • Adds sso_auth event to logs specifically for forced re-authentication events in SAMLSAMLA standard for passing signed identity assertions between a sign-in provider and an application. It lets a user sign in through their organization's identity provider./OIDC flows, providing visibility for compliance and security reviews.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds JumpCloud MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. Server (GA) to connect compatible AI tools to JumpCloud using the Model Context ProtocolModel Context ProtocolModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface., enabling interaction with org data and admin capabilities from AI tools and code editors.
  • Adds JumpCloud Access Requests with configurable approval flows, request management in the Admin Portal, tracking via Directory Insights, and webhook integration with ticketing systems such as JIRA.
  • Adds Shadow AI Dashboard — a dedicated UI for managing generative AI app usage, with insights into AI adoption, usage by department, and most active users — under the renamed AI & SaaS ManagementSaaS ManagementA category of software that tracks, governs, and optimises an organisation's portfolio of SaaS subscriptions, covering spend, access, and licence utilisation from a single control plane. product.
  • Adds Asset Manager role in the Admin Portal, providing dedicated access to Asset Management without granting broader Admin Portal access.
+19 more show less
  • Adds support for uploading and deploying IPA-based app packages to iOS and iPadOS devices via the JumpCloud Private Repository.
  • Adds Service Account for APIs, enabling API service accounts that are not tied to individual Admin accounts.
  • Adds Custom Admin Roles for direct organizations, enabling granular role configuration based on specific responsibilities.
  • Adds AI Commands Builder (GA) to create and refine commands for Windows, macOS, and Linux devices using natural-language prompts within the Admin Portal.
  • Adds ForceAuthn=true (SAML) and prompt=login (OIDC) support so applications can trigger step-up authentication without ending the user's existing session.
  • Adds Authentication Methods References (AMR) in SAML assertions and OIDC ID tokens to supply MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. proof to connected applications.
  • Adds Apple Business Manager as a new service discovery method for Account-driven Enrollment in Apple MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies..
  • Adds source IP tracking for incoming LDAPLDAPA protocol for querying and modifying directory services, standardized under RFC 4511, commonly used to look up users, groups, and attributes in corporate identity stores like Active Directory. requests for enhanced security and traceability.
  • Adds file upload capability (Invoice field) and File Upload custom field type to Asset Management asset records.
  • Adds Disk Size field under Hardware Specifications and Local Device Accounts under Device Management in the Asset Detail page.
  • Adds asset cloning to streamline creation of assets with shared hardware configurations, automatically clearing unique fields to prevent duplicates.
  • Adds User Group Exclusion from JumpCloud Protect Unified MFA Push enforcement, allowing excluded groups to use TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip. only.
  • Adds new Windows Personalization policies: Taskbar Customization, Screen Saver Configuration, and Desktop and Lock Screen Settings.
  • Adds faster reconnect for EAP connections by allowing recently authenticated devices to skip the full security handshake using a secure cached session credential.
  • Restricts JumpCloud LDAP TLS to five modern cipher suites: TLS_ECDHE_RSA_WITH_AES_128_GCM_SHA256, TLS_ECDHE_ECDSA_WITH_AES_128_GCM_SHA256, TLS_ECDHE_RSA_WITH_AES_256_GCM_SHA384, TLS_ECDHE_ECDSA_WITH_AES_256_GCM_SHA384, and TLS_RSA_WITH_AES_128_GCM_SHA256, removing CBC-mode and SHA-1SHA-1A cryptographic hash function standardized by NIST that produces a 160-bit digest; tools use it for file integrity checks, though it is now considered cryptographically broken and unsuitable for collision-resistant applications./SHA-256SHA-256A cryptographic hash function that turns data into a fixed 256-bit fingerprint. It is used to check integrity; hashing does not encrypt the original data. suites.
  • Adds SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally. License Management (GA) for automated license tracking, usage optimization, and cost management via SaaS application connectors.
  • Adds Customer Portal for tracking and managing support cases, including sortable columns for Last Modified Date, Account, and Requester, plus attachment uploads and inline ticket editing.
  • Expands Directory Insights Application Name and Resource Name filters into sub-categories: Application Display Name, Application Display Label, Resource Display Name, and Resource Display Label.
  • Adds Windows 11 25H2 support to JumpCloud Windows management components.
  • Adds sso_auth event to logs specifically for forced re-authentication events in SAMLSAMLA standard for passing signed identity assertions between a sign-in provider and an application. It lets a user sign in through their organization's identity provider./OIDC flows, providing visibility for compliance and security reviews.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds JumpCloud MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. Server (GA) to connect compatible AI tools to JumpCloud using the Model Context ProtocolModel Context ProtocolModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface., enabling interaction with org data and admin capabilities from AI tools and code editors.
  • Adds JumpCloud Access Requests with configurable approval flows, request management in the Admin Portal, tracking via Directory Insights, and webhook integration with ticketing systems such as JIRA.
  • Adds Shadow AI Dashboard — a dedicated UI for managing generative AI app usage, with insights into AI adoption, usage by department, and most active users — under the renamed AI & SaaS ManagementSaaS ManagementA category of software that tracks, governs, and optimises an organisation's portfolio of SaaS subscriptions, covering spend, access, and licence utilisation from a single control plane. product.
  • Adds Asset Manager role in the Admin Portal, providing dedicated access to Asset Management without granting broader Admin Portal access.
+14 more show less
  • Adds support for uploading and deploying IPA-based app packages to iOS and iPadOS devices via the JumpCloud Private Repository.
  • Adds Service Account for APIs, enabling API service accounts that are not tied to individual Admin accounts.
  • Adds Custom Admin Roles for direct organizations, enabling granular role configuration based on specific responsibilities.
  • Adds AI Commands Builder (GA) to create and refine commands for Windows, macOS, and Linux devices using natural-language prompts within the Admin Portal.
  • Adds ForceAuthn=true (SAML) and prompt=login (OIDC) support so applications can trigger step-up authentication without ending the user's existing session.
  • Adds Authentication Methods References (AMR) in SAML assertions and OIDC ID tokens to supply MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. proof to connected applications.
  • Adds Apple Business Manager as a new service discovery method for Account-driven Enrollment in Apple MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies..
  • Adds source IP tracking for incoming LDAPLDAPA protocol for querying and modifying directory services, standardized under RFC 4511, commonly used to look up users, groups, and attributes in corporate identity stores like Active Directory. requests for enhanced security and traceability.
  • Adds file upload capability (Invoice field) and File Upload custom field type to Asset Management asset records.
  • Adds asset cloning to streamline creation of assets with shared hardware configurations, automatically clearing unique fields to prevent duplicates.
  • Adds User Group Exclusion from JumpCloud Protect Unified MFA Push enforcement, allowing excluded groups to use TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip. only.
  • Adds new Windows Personalization policies: Taskbar Customization, Screen Saver Configuration, and Desktop and Lock Screen Settings.
  • Adds SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally. License Management (GA) for automated license tracking, usage optimization, and cost management via SaaS application connectors.
  • Adds Customer Portal for tracking and managing support cases, including sortable columns for Last Modified Date, Account, and Requester, plus attachment uploads and inline ticket editing.

5 more in Everything

└──▷ BREAKING ON UPGRADE
  • !JumpCloud LDAP no longer accepts connections using CBC-mode or SHA-1/SHA-256 integrity cipher suites (including RSARSAThe oldest widely used public-key algorithm, for both encryption and signatures, with security resting on factoring large numbers. A large quantum computer would break it. and ECDHE key exchange variants); only the five explicitly listed GCM cipher suites are permitted. Any LDAP client configured to use only the removed suites will fail to connect.
Release Notes 2026 (2026-09-21) Blog Stable

JumpCloud adds Password Vault across Admin/User Portal, mobile, and browser extension, plus GA Access Risk Detection and Agent Identities.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Register a new AI agent and scope its OAuth credentials to require credentials in the request body — needed when integrating with platforms like Amazon Bedrock AgentCore that enforce client_secret_post.
📍In the Admin Portal: Go to Identity Management › Agents › + Add. Fill in agent name, owner, and company email on the Details tab. On the Authentication tab, generate OAuth client credentials, then set the client authentication type to Client secret post to send the client ID and secret in the request body.
Surface which SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally. apps have locally running MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. servers and connect them to AI Gateway for centralized governance.
📍In the Admin Portal: Go to Access › AI & SaaS Management › Applications. Review the MCP Connection column for apps showing Discovered. Click the status for a target app, review discovered MCP servers, then click Add MCP Server to register them in AI Gateway.
  • Adds MCP Connection column (showing Discovered, Connected, or Not Available) in Access > AI & SaaS Management > Applications to surface MCP server status and enable adding servers to AI Gateway without switching tools.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Launches Password Vault in the JumpCloud Protect mobile app, letting users view secrets, autofill credentials into mobile apps, and autofill in the mobile browser using Push/TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip.-authenticated sessions.
  • Launches Password Vault in the Admin Portal, enabling admins to enable the vault, assign access to users and user groups, manage shared credentials and folders with Connect/View/Edit/Manage permissions, and view audit logs in Directory Insights.
  • Launches Password Vault in the User Portal, letting users manage and share credentials and website applications, bulk import via CSV or Excel, and group items in personal and shared folders.
  • Launches Password Vault Browser Extension via JumpCloud Go, supporting autofill of Non-SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. website passwords and 2FA codes, credential saving on login/update/signup, and secret viewing — without requiring device registration.
+8 more show less
  • Promotes Access Risk Detection to general availability, with behavior-based detection (unusual locations, impossible travel, unfamiliar devices/browsers, dormant accounts, abnormal login volumes), automatic MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone.-based resolution, intelligent suppression for VPNVPNA virtual private network, an encrypted tunnel between devices or networks. It carries traffic across another network as though the endpoints had a private connection./gateway logins, customizable risk factor thresholds, and email/Slack/webhook notifications.
  • Introduces Agent Identities (Public Preview) under Identity Management > Agents to register and govern non-human AI agent accounts with OAuth-based authentication, group-based application access, and full lifecycle management independent of human user accounts.
  • Adds a Dashboard tab to AI Gateway with org-wide MCP server usage metrics, tool call volume, error rates, trend charts, and per-server drill-down.
  • Adds client authentication type configuration on an agent's Authentication tab, supporting Client secret basic (credentials in authorization header, default) and Client secret post (credentials in request body) for OAuth token requests.
  • Adds Knowledge Base Uploads for AI Assistant, supporting PDF, TXT, DOCX, and MD files (10 MB max per file, 25 files max per org) in a private org-scoped knowledge base, with source attribution in chat responses.
  • Enables updating an existing application's binary in JumpCloud's Private Repository via Update Version on the application's Details tab, deploying to already-associated devices without re-creating the application or reassigning device groups.
  • Expands SCIMSCIMA standard for creating, updating and removing user accounts and groups across applications. It helps an identity provider keep access in sync when people join, change roles or leave. group management with a dedicated mapping UI supporting name overrides, extended schema mapping, side-by-side object transformation previews, flexible custom validation with selectable test users and groups, and automatic sync logging to Directory Insights.
  • Adds alert trigger conditions based on event frequency to Insights Rules, allowing alerts to fire only when defined thresholds are met rather than on every individual event.
  • Adds MCP Connection column (showing Discovered, Connected, or Not Available) in Access > AI & SaaS Management > Applications to surface MCP server status and enable adding servers to AI Gateway without switching tools.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Launches Password Vault in the JumpCloud Protect mobile app, letting users view secrets, autofill credentials into mobile apps, and autofill in the mobile browser using Push/TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip.-authenticated sessions.
  • Launches Password Vault in the Admin Portal, enabling admins to enable the vault, assign access to users and user groups, manage shared credentials and folders with Connect/View/Edit/Manage permissions, and view audit logs in Directory Insights.
  • Launches Password Vault in the User Portal, letting users manage and share credentials and website applications, bulk import via CSV or Excel, and group items in personal and shared folders.
  • Launches Password Vault Browser Extension via JumpCloud Go, supporting autofill of Non-SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. website passwords and 2FA codes, credential saving on login/update/signup, and secret viewing — without requiring device registration.
+2 more show less
  • Promotes Access Risk Detection to general availability, with behavior-based detection (unusual locations, impossible travel, unfamiliar devices/browsers, dormant accounts, abnormal login volumes), automatic MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone.-based resolution, intelligent suppression for VPNVPNA virtual private network, an encrypted tunnel between devices or networks. It carries traffic across another network as though the endpoints had a private connection./gateway logins, customizable risk factor thresholds, and email/Slack/webhook notifications.
  • Introduces Agent Identities (Public Preview) under Identity Management > Agents to register and govern non-human AI agent accounts with OAuth-based authentication, group-based application access, and full lifecycle management independent of human user accounts.

6 more in Everything

Subscribe to New Feature and Product Update Announcements (2026-09-21) Blog Stable

JumpCloud adds RSS feed subscriptions for release notes, product updates, agent notes, and bug fix announcements.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Subscribe to only the product and agent release notes RSS feed to track new JumpCloud capabilities in your RSS reader.
📍1. Go to the JumpCloud Release Notes and Bug Fixes page. 2. Click Subscribe to Release Notes and Bug Fixes RSS Feed. 3. Open your RSS reader and add the returned feed URL to begin receiving product release notes and agent release notes automatically.
What’s New: Password Manager Desktop App (2026-09-21) Blog LatestStable

JumpCloud Password Manager Desktop App adds EU region support, raised item limits, auto-lock scheduling, ARMARMA family of RISC processor architectures designed by Arm Holdings and licensed to chip makers worldwide; tools target it to run natively on mobile, embedded, and increasingly server and desktop hardware. build for Apple Silicon, and more.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Onboard EU-region users so their vault data stays within the EU — select the region before activating to ensure data residency compliance.
📍On the Password Manager activation page, select EU as your region before entering your activation code, or on the browser extension pairing page, choose the EU region from the region selector.
  • Introduces an auto-lock schedule with a default of 15 minutes of inactivity and a configurable maximum of 1440 minutes (24 hours).⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds Auto Lock setting that locks the Password Manager app automatically when the device is locked.
  • Scales item storage capacity to up to 13,000 items with autofill enabled, up to 5,000 items per shared folder, and up to 30,000 items when autofill is not used.
  • Adds a Move All items button in shared folders to move all items from a folder in a single action.
  • Introduces paginated items list displaying 100 items per page.
+14 more show less
  • Supports installation and update of the Password Manager desktop app via a new command template v.2.0.0 on Windows and macOS, eliminating the need for ongoing admin permissions after the initial run.
  • Extends Password Manager support to the EU region, with region selection available on the activation page (desktop app) and pairing page (browser extension).
  • Releases a native ARM build for macOS devices with M-series silicon (M1 and above), available on the Downloads page with automatic update for eligible devices.
  • Adds ability to delete multiple folders at once.
  • Enables admins to create and update shared folders, and view and assign users or user groups to any shared folder.
  • Adds item archiving functionality, allowing users to view, delete, and restore archived items.
  • Adds a Copy button to the Password History area.
  • Captures shared folder events and syncs them to Directory Insights in the JumpCloud Admin Portal.
  • Releases an RPM build for Linux (x64 Red Hat distributions such as Fedora) with auto-update support, in addition to the existing Debian build.
  • Adds What's New link under Settings > General so users can access a running list of desktop app updates.
  • Enhances CSV export for the legacy Password Manager with new data fields to support migration into JumpCloud Vault.
  • Enables users to sort items by nicknames.
  • Improves shared item management with multi-select for items and sorting by various criteria.
  • Adds Download Logs and Keyboard Shortcuts options under Settings.
  • Introduces an auto-lock schedule with a default of 15 minutes of inactivity and a configurable maximum of 1440 minutes (24 hours).⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds Auto Lock setting that locks the Password Manager app automatically when the device is locked.
  • Adds a Move All items button in shared folders to move all items from a folder in a single action.
  • Supports installation and update of the Password Manager desktop app via a new command template v.2.0.0 on Windows and macOS, eliminating the need for ongoing admin permissions after the initial run.
  • Extends Password Manager support to the EU region, with region selection available on the activation page (desktop app) and pairing page (browser extension).
+9 more show less
  • Releases a native ARM build for macOS devices with M-series silicon (M1 and above), available on the Downloads page with automatic update for eligible devices.
  • Adds ability to delete multiple folders at once.
  • Enables admins to create and update shared folders, and view and assign users or user groups to any shared folder.
  • Adds item archiving functionality, allowing users to view, delete, and restore archived items.
  • Adds a Copy button to the Password History area.
  • Captures shared folder events and syncs them to Directory Insights in the JumpCloud Admin Portal.
  • Releases an RPM build for Linux (x64 Red Hat distributions such as Fedora) with auto-update support, in addition to the existing Debian build.
  • Adds What's New link under Settings > General so users can access a running list of desktop app updates.
  • Adds Download Logs and Keyboard Shortcuts options under Settings.

5 more in Everything

└──▷ BREAKING ON UPGRADE
  • !The .appimage Linux build is no longer supported as of April 18, 2024; users must migrate to the RPM or Debian builds.
Was this useful?

Auth0

Sources Blog post → announced · Sep 18, 2026 Blog Rank Why this rank top features scored 85/78/76, +10 product media, +6 runnable example — 95.67 How ranking works →

Secure users, AI agents, and more with Auth0, an easy-to-implement, scalable, and adaptable authentication and authorization platform.

Auth0 shipped three Early Access/beta capabilities this window: an FGA Permissions Index for constant-time permission checks, an Agent Gateway for governing AI agent tool calls, and an Identity Conversion Suite adding Anonymous Sessions and auth-flow A/B testing.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 4 FEATURES 4 OF 4 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 01
Anonymous Sessions for guest identity tracking new 85

Issues OAuth 2.0OAuth 2.0An open authorization framework (IETF RFC 6749) that lets a service grant limited API access to a third party without sharing credentials, giving cyber tools a standard way to authenticate against external platforms.-compliant access tokensaccess tokensA credential that authorizes a client to use a service or API. It usually has limited permissions and a lifetime; anyone holding it may be able to use those permissions. to unauthenticated guest users, enabling guest checkout, cart tracking, pre-login personalization, and secure API access without authentication.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ The anonymous session token is passed to the Auth0 /authorize endpoint on sign-up or login, making all prior anonymous activity available inside Actions for custom handling, with anonymous metadata transferred seamlessly into the authenticated profile. Available in Early Access to Auth0 Enterprise plan customers.

Trigger an anonymous session and obtain an OAuth 2OAuth 2An open authorization framework (RFC 6749) that lets a service grant limited access to a user's resources on another service without exposing credentials, giving cyber tools a standard way to authenticate against third-party APIs..0 access tokenaccess tokenA credential that authorizes a client to use a service or API. It usually has limited permissions and a lifetime; anyone holding it may be able to use those permissions. for a guest user before they authenticate, so cart and wishlist metadata is tracked from first interaction.
📍Pass referral IDs, shopping cart IDs, or wishlist URLs to Auth0 at the start of the guest session. Auth0 issues a session token and an access token; use the access token to authorize API calls normally. When the guest signs up or logs in, include the session token in the request to the /authorize endpoint so Auth0 transfers all anonymous activity to the authenticated profile.
— Names `/authorize` endpoint and gives a runnable step sequence launch-20260918-8a7c8e15
02 02
Experiment Center for auth flow A/B testing new 78

Splits authentication traffic into variants (by percentage or segment) for capabilities such as passkeyspasskeysPhishing-resistant sign-in credentials built on FIDO2/WebAuthn: the device holds a private key and proves possession instead of the user typing a password. They sync across a user's devices., MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. steps, signup form copy, and branding themes across Universal Login (including Auth0 Custom Universal Login / ACUL), Forms, Actions, and Page Templates, without changing application code. Captures exposure, success rate, and time-to-auth per variant and segment as dedicated experiment events, enabling data-backed promotion of winning flows. Available in Early Access to Auth0 Enterprise plan customers.

A/B test a passkeypasskeyPhishing-resistant sign-in credentials built on FIDO2/WebAuthn: the device holds a private key and proves possession instead of the user typing a password. They sync across a user's devices.-first signup flow against the standard flow for warm visitors (those with saved cart items) to measure completion rate and time-to-auth before a full rollout.
📍In Experiment Center, define a feature flag for the passkey signup variant, assign the saved cart segment to the passkey variant and the remainder to control, then monitor the dedicated experiment events for exposure, success rate, and time-to-auth split by segment before promoting the winner tenant-wide.
— Clear UI-based starting point via feature flags and segments launch-20260918-8a7c8e15
03 03
Agent Gateway for AI agent tool-call authorization new 76

New beta control plane that evaluates tool-call authorization in real time using agent, user, and organization context before any external tool is invoked. Integrates with Auth0 Token Vault and on-behalf-of token exchangetoken exchangeAn OAuth 2.0 protocol extension (RFC 8693) that lets a service trade one token for another with different scopes, audience, or type, enabling delegation and impersonation across service boundaries without re-authenticating the user. to broker short-lived, user-scoped tokens, keeping raw credentials out of LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code. prompts, context windowscontext windowsThe maximum amount of text, counted in tokens, a model can take in and produce in one request. Everything the model knows about the task has to fit inside it., and logs. Supports organization-aware, multi-tenant routing that isolates tool access and connected credentials (e.g. Slack, Microsoft Teams) per customer tenant, includes an agent kill switch to disable an agent or restrict its access across a tenant, and logs every agent tool call to tenant logs with optional streaming export to SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents. services. Exposes Agent Gateway APIs for early integration during the beta, with documentation and integration guides.

— Names integrations and controls but no exact endpoint or flag launch-20260918-5ac20484
04 04
FGA Permissions Index for constant-time access checks new 68

Precomputes every access implied by chosen relations (e.g. can_view) on an object type and streams the results directly into the customer's own database, enabling a simple SQL JOIN instead of runtime graph traversal for permission-filtered queries. Supports indexing high-cardinality relations so lookups on indexed relations become constant-time, while non-indexed relations continue to be evaluated by FGA in real time.

— Explains mechanism and scope but gives no API or config surface launch-20260918-135298a6
announced · Sep 18, 2026 Blog

Auth0 FGA Permissions Index (Early Access) precomputes all user permissions and streams them into your own database for constant-time SQL JOIN-based access checks at scale.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Introduces FGA Permissions Index (Early Access), which precomputes every access implied by chosen relations (e.g.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ can_view) on an object type and streams the results directly into the customer's own database, enabling a simple SQL JOIN instead of runtime graph traversal for permission-filtered queries.
  • Supports indexing high-cardinality relations on an object type so that permission lookups on indexed relations become constant-time, while non-indexed relations continue to be evaluated by FGA in real time.

2 details in Everything

announced · Sep 18, 2026 Blog

Auth0 Agent Gateway (beta) adds a CIAM control plane for governing customer-facing AI agents across multi-tenant SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally. products.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • New Auth0 Agent Gateway beta provides a centralized policy-enforcement layer that evaluates tool-call authorization in real time using agent, user, and organization context before any external tool is invoked.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Integrates with Auth0 Token Vault and on-behalf-of token exchangetoken exchangeAn OAuth 2.0 protocol extension (RFC 8693) that lets a service trade one token for another with different scopes, audience, or type, enabling delegation and impersonation across service boundaries without re-authenticating the user. to broker short-lived, user-scoped tokens for internal and external APIs — keeping raw credentials out of LLMLLMA large language model, a neural network trained on large amounts of text to recognize patterns and generate language. Applications use it for tasks such as answering, summarizing and writing code. prompts, context windowscontext windowsThe maximum amount of text, counted in tokens, a model can take in and produce in one request. Everything the model knows about the task has to fit inside it., and logs.
  • Supports organization-aware, multi-tenant routing that isolates tool access and connected credentials (e.g. Slack, Microsoft Teams) per customer tenant, preventing cross-tenant data exposure.
  • Provides an agent kill switch — centralized administrative controls to disable an agent or restrict its access to downstream services and tools across a tenant or deployment.
  • Logs every agent tool call to tenant logs with optional streaming export to SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents. services for enterprise auditability.
+1 more show less
  • Exposes Auth0 Agent Gateway APIs for early integration during the beta, with accompanying documentation, integration guides, and examples.

6 details in Everything

announced · Sep 18, 2026 Blog

Auth0 launches Identity Conversion Suite (Early Access) with Anonymous Sessions and Experiment Center for auth flow A/B testing⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Trigger an anonymous session and obtain an OAuth 2.0OAuth 2.0An open authorization framework (IETF RFC 6749) that lets a service grant limited API access to a third party without sharing credentials, giving cyber tools a standard way to authenticate against external platforms. access tokenaccess tokenA credential that authorizes a client to use a service or API. It usually has limited permissions and a lifetime; anyone holding it may be able to use those permissions. for a guest user before they authenticate, so cart and wishlist metadata is tracked from first interaction.
📍Pass referral IDs, shopping cart IDs, or wishlist URLs to Auth0 at the start of the guest session. Auth0 issues a session token and an access token; use the access token to authorize API calls normally. When the guest signs up or logs in, include the session token in the request to the /authorize endpoint so Auth0 transfers all anonymous activity to the authenticated profile.
A/B test a passkeypasskeyPhishing-resistant sign-in credentials built on FIDO2/WebAuthn: the device holds a private key and proves possession instead of the user typing a password. They sync across a user's devices.-first signup flow against the standard flow for warm visitors (those with saved cart items) to measure completion rate and time-to-auth before a full rollout.
📍In Experiment Center, define a feature flag for the passkey signup variant, assign the saved cart segment to the passkey variant and the remainder to control, then monitor the dedicated experiment events for exposure, success rate, and time-to-auth split by segment before promoting the winner tenant-wide.
  • Passes the anonymous session token to the Auth0 /authorize endpoint on sign-up or login, making all prior anonymous activity available inside Actions for custom handling.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds Anonymous Sessions: issues OAuth 2OAuth 2An open authorization framework (RFC 6749) that lets a service grant limited access to a user's resources on another service without exposing credentials, giving cyber tools a standard way to authenticate against third-party APIs..0-compliant access tokensaccess tokensA credential that authorizes a client to use a service or API. It usually has limited permissions and a lifetime; anyone holding it may be able to use those permissions. to unauthenticated guest users, enabling guest checkout, cart tracking, pre-login personalization, and secure API access without authentication — with seamless transfer of anonymous metadata into the authenticated profile on sign-up or login.
  • Adds Experiment Center: split authentication traffic into variants (by percentage or segment) for capabilities such as passkeyspasskeysPhishing-resistant sign-in credentials built on FIDO2/WebAuthn: the device holds a private key and proves possession instead of the user typing a password. They sync across a user's devices., MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. steps, signup form copy, and branding themes — across Universal Login (including Auth0 Custom Universal Login / ACUL), Forms, Actions, and Page Templates — without changing application code.
  • Experiment Center captures exposure, success rate, and time-to-auth per variant and segment as dedicated experiment events, enabling data-backed promotion of winning flows.
  • Both Anonymous Sessions and Experiment Center are available in Early Access to Auth0 Enterprise plan customers.

5 details in Everything

Was this useful?
◆  Data Security & DLP

BigID

Sources Blog / feed → Meet AgentIQ: Run Your Entire Data & AI Program From a Prompt (2026-09-21) Blog Rank Why this rank top features scored 70/60/50 — 60 How ranking works →

BigID launched AgentIQ, a new agentic automation layer that lets teams run data and AI security workflows via natural language, exposing discovery, classification, remediation and governance to external AI assistants through MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface., alongside pre-built agents, human-in-the-loophuman-in-the-loopHuman in the loop, involving a person in an otherwise automated workflow. The person can review, approve, correct or supply information before work continues. approval controls, shadow AI discovery, and automated reporting across 200+ integrations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Secure your data everywhere—cloud, on-prem, and AI. BigID delivers DSPM and AI governance trusted by Fortune 500 enterprises.

BigID launched AgentIQ, a new agentic automation layer that lets teams run data and AI security workflows via natural language, exposing discovery, classification, remediation and governance to external AI assistants through MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface., alongside pre-built agents, human-in-the-loophuman-in-the-loopHuman in the loop, involving a person in an otherwise automated workflow. The person can review, approve, correct or supply information before work continues. approval controls, shadow AI discovery, and automated reporting across 200+ integrations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 6 FEATURES 0 OF 6 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
AgentIQ agentic automation layer via MCP new 70

AgentIQ lets practitioners describe data and AI security workflows in natural language and have an agent investigate, prioritize, report, and take action across BigID's discovery, classification, access intelligence, and remediation capabilities.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ It exposes BigID's operational capabilities — discover, classify, investigate, remediate, govern, and fulfill — to external AI assistants (Claude, ChatGPT, Copilot, Gemini, and custom enterprise agents) through MCP and APIs, so teams can operate BigID from whichever AI interface they already use.

— Names MCP and specific AI assistants but no exact commands or endpoints. Meet AgentIQ: Run Your Entire Data & AI Prog…
02 00
Human-in-the-loop controls for agent actions new 60

Revoking access, quarantining data, and deleting records can require human approval, with all agent actions logged and attributable to the requesting user's permissions enforced at the API and MCP layer.

— Explains approval and logging mechanism but no config path given. Meet AgentIQ: Run Your Entire Data & AI Prog…
thinner coverage below
03 00
Pre-built AgentIQ agents for common workflows new 50

Ships pre-built agents for sensitive data remediation, AI governance enforcement, access exposure analysis, and privacy operations, enabling teams to start without building custom agent workflows.

— Names agent categories but no configuration detail. Meet AgentIQ: Run Your Entire Data & AI Prog…
04 00
Shadow AI and sanctioned AI discovery workflows new 45

Agents can inventory AI systems across the environment, map the data each system consumes, and flag policy violations.

— Describes function but no scope or trigger mechanism. Meet AgentIQ: Run Your Entire Data & AI Prog…
05 00
On-demand and scheduled AgentIQ reporting new 45

Supports on-demand and scheduled report generation across board-level summaries, auditor evidence packs, and recurring security team updates, driven by metadata, scan findings, and activity data.

— Lists report types but no format or scheduling detail. Meet AgentIQ: Run Your Entire Data & AI Prog…
06 00
200+ integrations through AgentIQ framework new 35

Covers more than 200 integrations spanning on-premises, cloud, SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally., and API sources through the AgentIQ framework.

— Gives a number but no list of named integrations. Meet AgentIQ: Run Your Entire Data & AI Prog…
Meet AgentIQ: Run Your Entire Data & AI Program From a Prompt (2026-09-21) Blog LatestStable

BigID launches AgentIQ, enabling agentic automation of data security workflows via MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. through Claude, Copilot, ChatGPT, Gemini, and custom enterprise agents.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Introduces AgentIQ, a new agentic automation layer that lets practitioners describe data and AI security workflows in natural language and have an agent investigate, prioritize, report, and take action across BigID's discovery, classification, access intelligence, and remediation capabilities.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Exposes BigID's operational capabilities — discover, classify, investigate, remediate, govern, and fulfill — to external AI assistants (Claude, ChatGPT, Copilot, Gemini, and custom enterprise agents) through MCP and APIs, so teams can operate BigID from whichever AI interface they already use.
  • Ships pre-built agents for sensitive data remediation, AI governance enforcement, access exposure analysis, and privacy operations, enabling teams to start without building custom agent workflows.
  • Supports human-in-the-loophuman-in-the-loopHuman in the loop, involving a person in an otherwise automated workflow. The person can review, approve, correct or supply information before work continues. controls where revoking access, quarantining data, and deleting records can require human approval, with all agent actions logged and attributable to the requesting user's permissions enforced at the API and MCP layer.
  • Enables shadow AI and sanctioned AI discovery workflows: agents can inventory AI systems across the environment, map the data each system consumes, and flag policy violations.
+2 more show less
  • Supports on-demand and scheduled report generation across board-level summaries, auditor evidence packs, and recurring security team updates, driven by metadata, scan findings, and activity data.
  • Covers more than 200 integrations spanning on-premises, cloud, SaaSSaaSSoftware delivered over the internet and run on the provider's infrastructure, so the buyer pays a subscription instead of installing or operating anything locally., and API sources through the AgentIQ framework.

7 details in Everything

Was this useful?
Governance, Risk & Compliance
◆  Continuous Control Monitoring (CCM)

RegScale

Sources Blog / feed → CLI 6.48.0 (2026-09-21) Blog Rank Why this rank top features scored 60/50 — 55 How ranking works →

RegScale's CLI 6.48.0 release adds finer control over JCAM import scoping and enriches Wiz issue sync with additional context fields.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

RegScale is a compliance and security governance platform that automates regulatory assessment, documentation, and continuous monitoring across organizations.

RegScale's CLI 6.48.0 release adds finer control over JCAM import scoping and enriches Wiz issue sync with additional context fields.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 2 FEATURES 0 OF 2 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
Repeatable plan ID argument for JCAM imports new 60

JCAM import commands gain a repeatable plan id argument that lets a sync run be scoped to specific plans without editing the configured filter.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

— Names exact flag but no example of use CLI 6.48.0 (2026-09-21)
thinner coverage below
02 00
Enriched Wiz issue sync metadata improved 50

Wiz issue sync now imports issue notes, the affected resource's cloud console link, project business impact, policy type, and linked service tickets.

— Lists new fields but no config or command shown CLI 6.48.0 (2026-09-21)
CLI 6.48.0 (2026-09-21) Blog LatestStable

RegScale CLI 6.48.0 adds repeatable JCAM plan ID scoping and enriched Wiz issue sync with notes, links, and metadata.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

  • Adds a repeatable plan id argument to JCAM import commands so a sync run can be scoped to specific plans without editing the configured filter.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Wiz issue sync now imports issue notes, the affected resource's cloud console link, project business impact, policy type, and linked service tickets.
  • Adds a repeatable plan id argument to JCAM import commands so a sync run can be scoped to specific plans without editing the configured filter.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

1 more in Everything

Was this useful?
Govern
◆  AI/LLM Security

ai-safe2-framework

Sources Release notes → 2026_09_20_CLI_0.7.0 Notes Rank Why this rank top features scored 95/95/55 — 81.67 How ranking works →

AI SAFE² CLI 0.7.0 introduces two new safe2 evidence subcommands for correlating multi-harness agent evidence and detecting agent configuration drift, backed by new contract schemas and hardened evidence-handling protections.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Ai-Safe2-Framework provides governance, risk, and compliance controls for securing agentic AI systems and non-human identities.

AI SAFE² CLI 0.7.0 introduces two new safe2 evidence subcommands for correlating multi-harness agent evidence and detecting agent configuration drift, backed by new contract schemas and hardened evidence-handling protections.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 3 FEATURES 0 OF 3 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
`safe2 evidence truth` correlation command new 95

Adds safe2 evidence truth to correlate harness evidence, revision-bound task receipts, coverage, usage declarations, and completion claims across multiple agent harnessesagent harnessesThe software around an AI agent that manages its tools, execution loop, state and limits. It determines how the model's proposed actions are run and checked..⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Uses the new operational-truth-source-v1 contract schema to declare the task, revision, required evidence, receipt bindings, decision owner, assumptions, and exclusions, and the operational-truth-manifest-v1 schema to record the correlated result, decision gate (hold, review, ready_for_human_decision), conflicts, gaps, and evidence boundaries.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Supports --baseline, --output, --strict, and --card flags, with --card generating a human decision card alongside canonical JSON output from the same evaluated evidence record.

— Named command, schemas and flags with clear mechanism and behaviour. 2026_09_20_CLI_0.7.0
02 00
`safe2 evidence changes` drift detection command new 95

Adds safe2 evidence changes to run a bounded one-shot inventory of skills and recognized agent configuration locally or in CI, detecting additions, modifications, and removals. It recognizes skill packages rooted by SKILL.md, MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. configuration, OpenClaw configuration, and supported configuration beneath .codex, .claude, .hermes, and .openclaw, records results via the new change-monitor-v1 contract schema without exporting configuration contents, and enforces bounded traversal of up to 10,000 entries and 1 MB per tracked file by default. Also supports the --baseline, --output, --strict, and --card flags.

— Named command, schema, recognized paths and limits, fully actionable. 2026_09_20_CLI_0.7.0
thinner coverage below
03 00
Strict evidence-contract protections new 55

Adds strict evidence-contract protections applying to the evidence commands: duplicate-key rejection, non-finite-number rejection, symbolic-link and reparse-point guards, no-overwrite behavior, and exclusive complete-before-visible evidence publication.

— Lists distinct protections but no flags or commands to invoke them. 2026_09_20_CLI_0.7.0
2026_09_20_CLI_0.7.0 Notes LatestStable

AI SAFE² CLI 0.7.0 adds safe2 evidence truth and safe2 evidence changes to correlate multi-harness agent evidence and detect configuration drift.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch 2026_09_20_CLI_0.7.0 https://github.com/CyberStrategyInstitute/ai-safe2-framework.git
# already have the repo? check out this version:
$ git checkout 2026_09_20_CLI_0.7.0
  • Adds safe2 evidence truth command to correlate harness evidence, revision-bound task receipts, coverage, usage declarations, and completion claims across multiple agent harnessesagent harnessesThe software around an AI agent that manages its tools, execution loop, state and limits. It determines how the model's proposed actions are run and checked..⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds safe2 evidence changes command to run a bounded one-shot inventory of skills and recognized agent configuration locally or in CI, detecting additions, modifications, and removals.
  • Adds --baseline, --output, --strict, and --card flags to safe2 evidence subcommands for controlling baseline comparison, artifact output paths, strict-mode enforcement, and human-readable decision card generation.
  • Introduces operational-truth-source-v1 contract schema to declare the task, revision, required evidence, receipt bindings, decision owner, assumptions, and exclusions.
  • Introduces operational-truth-manifest-v1 contract schema to record the correlated result, decision gate (hold, review, ready_for_human_decision), conflicts, gaps, and evidence boundaries.
+5 more show less
  • Introduces change-monitor-v1 contract schema to record bounded agent-input inventory and change results without exporting configuration contents.
  • Recognizes skill packages rooted by SKILL.md, MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. configuration, OpenClaw configuration, and supported configuration beneath .codex, .claude, .hermes, and .openclaw during change monitoring.
  • Generates a human decision card (--card) alongside canonical JSON output from the same evaluated evidence record.
  • Enforces bounded traversal of up to 10,000 entries and 1 MB per tracked file by default during change monitoring.
  • Adds strict evidence-contract protections including duplicate-key rejection, non-finite-number rejection, symbolic-link and reparse-point guards, no-overwrite behavior, and exclusive complete-before-visible evidence publication.

10 details in Everything

Was this useful?

Docker Sandboxes

Sources Release notes → v0.45.0 Notes Rank Why this rank top features scored 85/83/80, +15 breaking change, +6 runnable example — 103.67 How ranking works →

Docker Sandboxes isolates applications in lightweight containers, providing secure, reproducible environments for development and deployment.

Docker Sandboxes v0.45.0 introduces v3 OCIOCIOpen Container Initiative standard, maintained by the Linux Foundation, defining the image format and runtime spec that container tools share so images built once run anywhere OCI-compliant. kits with composable workload and mixin roles and kit sets, adds sbx run --model support for OpenAI/Anthropic-compatible endpoints, and hardens sandbox network policy enforcement, alongside a wave of new CLI flags and several breaking changes to error handling and destructive-command confirmation.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 12 FEATURES 1 OF 12 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
New CLI flags across MCP, list, network and policy commands new 85

Adds sbx mcp add --disable-http2 to disable HTTP/2HTTP/2The second major version of the HTTP protocol, standardized by the IETF, adding multiplexed streams, header compression, and binary framing over a single TCP connection to cut latency and reduce overhead. for a remote MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. transport, working around servers whose HTTP/2 handling stalls long-lived streams.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ Adds --quiet (-q) flag to MCP server, secret, skill, template, volume, and policy-profile list commands for name-only output. Adds --protocol flag to local network allow rules so UDPUDPA network protocol that sends individual messages without guaranteeing delivery or order. Applications trade built-in reliability for less overhead and can add their own recovery. can be selected explicitly alongside the default TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it. coverage. Adds --created-via filter to sbx policy ls, which also now shows how each rule was created.

Allow outbound UDP (e.g. for a DNS-over-UDP workflow) in a sandbox network policy instead of the TCP-only default.
$ sbx policy allow network --protocol udp 8.8.8.8:53
Work around a remote MCP server that stalls long-lived HTTP/2 streams by forcing HTTP/1.1.
$ sbx mcp add --disable-http2 my-mcp-server https://mcp.example.com/sse
— Every flag named verbatim with two runnable examples v0.45.0
02 01
Breaking CLI behavior changes for removals and errors breaking 83

sbx mcp catalog has been removed; register a remote MCP server with sbx mcp add and then authorize it with sbx mcp auth instead. sbx secret rm now returns an error on stderr when the requested secret does not exist (previously silent). sbx mcp rm now returns an error when the requested MCP server is not registered (previously silent). Commands that remove resources now prompt for confirmation and return a non-zero exit code when the prompt is declined; use --force or --yes/-y in non-interactive workflows. sbx exec -d / --detach now fails immediately for unsupported detached execution instead of silently running in the foreground.

— Five named commands with exact before/after behaviour and workaround flags v0.45.0
03 00
Model provider support for `sbx run --model` new 80

sbx run --model now supports any OpenAI- or Anthropic-compatible endpoint configured via the new model.providers setting.

Route agent runs through a custom Anthropic-compatible endpoint instead of the built-in provider.
$ sbx run --model my-anthropic-proxy/claude-3-5-sonnet 'summarize the repo'
— Named setting plus a runnable example command v0.45.0
04 00
`sbx kit validate` rejects malformed API-key declarations new 78

sbx kit validate now rejects malformed API-key declarations, including invalid names, missing injection domains, invalid format placeholders, and Basic-auth usernames containing a colon, and warns on no-op or out-of-allowlist declarations.

— Specific validation cases named, no example run v0.45.0
05 00
v3 OCI kits with workload/mixin composition and kit sets new 73

v3 kits are OCI-based packages with separate workload and mixin roles that combine base environments, tools, configuration, credentials, network access, and agent instructions. Kit sets let authors combine a workload and mixins, pin component versions, and publish the result as a single OCI reference with its own capabilities, lifecycle hooks, instructions, and arguments. V3 kits can scope network access by HTTP method and path, declare install-phase network access and credential use, and specify where an agent reads shared skills. Multiple OAuth-backed agents can now be composed in the same sandbox with credentials scoped to the individual kits that request them. V3 workloads and mixins must be used together; they cannot be combined with v1 or v2 kits.

— Rich mechanism but no exact command shown to try it v0.45.0
06 00
OpenCode kit model variants and Copilot integration new 70

sbx run opencode --model now exposes the model's supported thinking levels as OpenCode variants, selectable with Ctrl+T. The OpenCode kit now configures GitHub Copilot from the account's stored GitHub credential, enabling Copilot models without a separate device login.

— Command and keybinding named, mechanism explained v0.45.0
07 00
Expanded JSON output for sandbox listing and inspection improved 60

Adds created_at, CPUCPUThe main general-purpose processor in a computer. It runs application instructions and coordinates work with memory and other hardware. limits, and memory limits fields to sbx ls --json and sbx inspect --json output.

— Exact fields and commands named but no example shown v0.45.0
thinner coverage below
08 00
Malformed network pattern validation in policy commands improved 55

sbx policy allow network, sbx policy deny network, --allow-network, and --deny-network now reject malformed patterns before saving them.

— Commands named but behaviour described in one line v0.45.0
09 00
Network policy enforcement hardening improved 55

Reverse-DNS lookups are now restricted to destination IPs already authorized by policy, closing a previous policy bypass. DNS resolution is blocked when no network rule permits it. Experimental outbound UDP now follows sandbox network policy.

— Clear before/after but no commands to act on v0.45.0
10 00
macOS distribution as signed Sbx.app bundle improved 50

macOS distributions now ship as a single signed Sbx.app bundle, with Homebrew and tarball PATH installs working via a symlink into the bundle.

— Packaging change described but no install steps given v0.45.0
11 00
Live update of global service secrets to running sandboxes improved 40

Adding, updating, or removing global service secrets now updates existing local sandboxes without a restart.

— Behaviour stated plainly with no named command v0.45.0
12 00
OAuth credential refresh coordination across sandboxes improved 35

OAuth refreshes are coordinated across sandboxes that share credentials, preventing simultaneous refreshes from forcing another sign-in.

— Describes benefit but no mechanism detail or surface v0.45.0
└──▷ BREAKING ON UPGRADE
  • !
    sbx mcp catalog has been removed; register a remote MCP server with sbx mcp add and then authorize it with sbx mcp auth instead.
  • !
    sbx secret rm now returns an error on stderr when the requested secret does not exist (previously silent).
  • !
    sbx mcp rm now returns an error when the requested MCP server is not registered (previously silent).
  • !
    V3 workloads and mixins must be used together; they cannot be combined with v1 or v2 kits.
  • !
    Commands that remove resources now prompt for confirmation and return a non-zero exit code when the prompt is declined; use --force or --yes/-y in non-interactive workflows.
  • !
    sbx exec -d / --detach now fails immediately for unsupported detached execution instead of silently running in the foreground.
v0.45.0 Notes LatestStable

Docker Sandboxes v0.45.0 ships v3 OCIOCIOpen Container Initiative standard, maintained by the Linux Foundation, defining the image format and runtime spec that container tools share so images built once run anywhere OCI-compliant. kits with workload/mixin composition, new model provider settings, and a raft of new CLI flags.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ GET THIS VERSION 
$ git clone --branch v0.45.0 https://github.com/docker/sbx-releases.git
# already have the repo? check out this version:
$ git checkout v0.45.0
└──▷ TRY IT
Route agent runs through a custom Anthropic-compatible endpoint instead of the built-in provider.
$ sbx run --model my-anthropic-proxy/claude-3-5-sonnet 'summarize the repo'
Allow outbound UDPUDPA network protocol that sends individual messages without guaranteeing delivery or order. Applications trade built-in reliability for less overhead and can add their own recovery. (e.g. for a DNS-over-UDP workflow) in a sandbox network policy instead of the TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it.-only default.
$ sbx policy allow network --protocol udp 8.8.8.8:53
Work around a remote MCPMCPModel Context Protocol, a standard for connecting AI applications to external tools and data sources. A compatible server exposes capabilities an assistant can use through a common interface. server that stalls long-lived HTTP/2HTTP/2The second major version of the HTTP protocol, standardized by the IETF, adding multiplexed streams, header compression, and binary framing over a single TCP connection to cut latency and reduce overhead. streams by forcing HTTP/1.1.
$ sbx mcp add --disable-http2 my-mcp-server https://mcp.example.com/sse
  • Adds sbx run --model support for any OpenAI- or Anthropic-compatible endpoint configured via the new model.providers setting.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds sbx mcp add --disable-http2 to disable HTTP/2 for a remote MCP transport, working around servers whose HTTP/2 handling stalls long-lived streams.
  • Adds --quiet (-q) flag to MCP server, secret, skill, template, volume, and policy-profile list commands for name-only output.
  • Adds --protocol flag to local network allow rules so UDP can be selected explicitly alongside the default TCP coverage.
  • Adds --created-via filter to sbx policy ls, which also now shows how each rule was created.
+16 more show less
  • Adds --force (and --yes/-y for sbx kit builder history rm) flags for non-interactive confirmation of destructive commands.
  • Adds created_at, CPUCPUThe main general-purpose processor in a computer. It runs application instructions and coordinates work with memory and other hardware. limits, and memory limits to sbx ls --json and sbx inspect --json output.
  • Adds sbx kit validate rejection of malformed API-key declarations, including invalid names, missing injection domains, invalid format placeholders, and Basic-auth usernames containing a colon, plus warnings for no-op or out-of-allowlist declarations.
  • Introduces v3 kits: OCI-based packages with separate workload and mixin roles that combine base environments, tools, configuration, credentials, network access, and agent instructions.
  • Introduces kit sets, letting authors combine a workload and mixins, pin component versions, and publish the result as a single OCI reference with its own capabilities, lifecycle hooks, instructions, and arguments.
  • V3 kits can scope network access by HTTP method and path, declare install-phase network access and credential use, and specify where an agent reads shared skills.
  • Multiple OAuth-backed agents can now be composed in the same sandbox with credentials scoped to the individual kits that request them.
  • Experimental outbound UDP now follows sandbox network policy.
  • Adding, updating, or removing global service secrets now updates existing local sandboxes without a restart.
  • OAuth refreshes are coordinated across sandboxes that share credentials, preventing simultaneous refreshes from forcing another sign-in.
  • Reverse-DNS lookups are now restricted to destination IPs already authorized by policy, closing a previous policy bypass.
  • DNS resolution is blocked when no network rule permits it.
  • sbx policy allow network, sbx policy deny network, --allow-network, and --deny-network now reject malformed patterns before saving them.
  • sbx run opencode --model now exposes the model's supported thinking levels as OpenCode variants, selectable with Ctrl+T.
  • The OpenCode kit now configures GitHub Copilot from the account's stored GitHub credential, enabling Copilot models without a separate device login.
  • macOS distributions now ship as a single signed Sbx.app bundle, with Homebrew and tarball PATH installs working via a symlink into the bundle.
  • Adds sbx run --model support for any OpenAI- or Anthropic-compatible endpoint configured via the new model.providers setting.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Introduces v3 kits: OCI-based packages with separate workload and mixin roles that combine base environments, tools, configuration, credentials, network access, and agent instructions.
  • Introduces kit sets, letting authors combine a workload and mixins, pin component versions, and publish the result as a single OCI reference with its own capabilities, lifecycle hooks, instructions, and arguments.
  • V3 kits can scope network access by HTTP method and path, declare install-phase network access and credential use, and specify where an agent reads shared skills.
  • Multiple OAuth-backed agents can now be composed in the same sandbox with credentials scoped to the individual kits that request them.
+2 more show less
  • Experimental outbound UDP now follows sandbox network policy.
  • OAuth refreshes are coordinated across sandboxes that share credentials, preventing simultaneous refreshes from forcing another sign-in.

14 more in Everything

└──▷ BREAKING ON UPGRADE
  • !sbx mcp catalog has been removed; register a remote MCP server with sbx mcp add and then authorize it with sbx mcp auth instead.
  • !sbx secret rm now returns an error on stderr when the requested secret does not exist (previously silent).
  • !sbx mcp rm now returns an error when the requested MCP server is not registered (previously silent).
  • !V3 workloads and mixins must be used together; they cannot be combined with v1 or v2 kits.
  • !Commands that remove resources now prompt for confirmation and return a non-zero exit code when the prompt is declined; use --force or --yes/-y in non-interactive workflows.
  • !sbx exec -d / --detach now fails immediately for unsupported detached execution instead of silently running in the foreground.
Was this useful?
Infrastructure
◆  Internet Services

Cloudflare

Sources Blog / feed → WAF - Enforce positive security with Application Profiles (2026-09-07) Blog Rank Why this rank top features scored 87/83/65, +15 breaking change, +6 runnable example — 99.33 How ranking works →

Connect, protect, and build everywhere.

Cloudflare's biggest shift this window is Application Profiles, a new positive-security enforcement layer for WAFWAFA reverse proxy that inspects HTTP/HTTPS traffic and blocks requests matching known attack patterns such as SQL injection or XSS, sitting between clients and a web application to filter malicious input before it reaches the server., alongside a sweeping expansion of the Workers AI model catalog and general availability for WAN Unified Routing; smaller improvements landed for Workflows, Hyperdrive, RPCRPCRemote Procedure Call, a protocol that lets a program execute a function on another machine as if it were local, giving cyber tools a structured way to communicate across processes or hosts without managing raw sockets. tracing, and Browser Rendering debugging.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ WHAT SHIPPED · 9 FEATURES 1 OF 9 FEATURES most completely described first
what's the number?

Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.

  • depth0–40 what it does and how it works or what changed
  • specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
  • actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point

Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.

01 00
New models added to Workers AI catalog new 87

Adds 15 new models to Workers AI: wan-3.0-prime (Alibaba text-to-video, 480P/720P/1080P, adaptive aspect ratio, configurable duration); h3 (multimodalmultimodalAble to work with more than one type of input or output, such as text, images and audio. A multimodal model can combine information across those formats. video generation, text-to-video, first/last frame image-to-video, reference-to-video, 768P/2K); h3-max (fast multimodal video generation, text-to-video, image-to-video, 480P/768P); gemini-omni-1.1-flash and gemini-omni-flash (multimodal video generation/editing with conversational controls and generated audio); gemini-3.8-flash (long-horizon software engineering, autonomous agents, enterprise workflows); gpt-6-astra (OpenAI, complex reasoning, coding, computer use, research, document creation); gpt-image-2.5-flare and gpt-image-2.5-sunburst (image generation/editing with low, medium, high, xhigh, max, auto quality settings); gpt-oss-120b and gpt-oss-20b (open-weight models for production and low-latency/local use respectively); union-alpha (multimodal, research/coding/agentic workflows); jev (TypeSafe structured evaluation, typed Noul, Choice, and Score questions with calibrated probabilities and confidence scores); pareto (Unbiased, runs multiple language models in parallel per request and synthesizes one response, text and vision inputs); and FLUX.2 [klein] 9B (9-billion-parameter image generation with multi-reference editing).⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

Generate a 1080P cinematic video from a text prompt using the new wan-3.0-prime model via the Workers AI REST APIREST APIA web API design style where clients interact with resources over HTTP using standard methods like GET, POST, and DELETE, giving cyber tools a predictable, language-agnostic way to integrate with external services..
$ curl https://api.cloudflare.com/client/v4/accounts/<ACCOUNT_ID>/ai/run/@cf/alibaba/wan-3.0-prime -H 'Authorization: Bearer <API_TOKEN>' -H 'Content-Type: application/json' -d '{"prompt": "A drone shot of a mountain range at sunrise", "resolution": "1080P"}'
Evaluate a model output against structured Choice and Score questions using the jev structured evaluation model.
$ curl https://api.cloudflare.com/client/v4/accounts/<ACCOUNT_ID>/ai/run/@cf/typesafe/jev -H 'Authorization: Bearer <API_TOKEN>' -H 'Content-Type: application/json' -d '{"state": "The response was accurate and concise.", "questions": [{"type": "Choice", "text": "Is the response helpful?"}, {"type": "Score", "text": "Rate response quality 1-5"}]}'
Generate a high-quality image with gpt-image-2.5-sunburst using the xhigh quality setting for detailed output.
$ curl https://api.cloudflare.com/client/v4/accounts/<ACCOUNT_ID>/ai/run/@cf/openai/gpt-image-2.5-sunburst -H 'Authorization: Bearer <API_TOKEN>' -H 'Content-Type: application/json' -d '{"prompt": "A photorealistic castle on a cliff at dusk", "quality": "xhigh"}'
— Names every model and spec, with runnable curl examples for some product docs
02 01
Application Profiles positive-security enforcement in WAF new 83

Introduces Application Profiles, a new layer for Cloudflare WAF that learns expected request structure and classifies traffic as conforming or non-conforming without blocking. The first profile type, Schema Profiles, models path variables, query parameters, headers, cookies, JSON bodies, and form-encoded bodies — including field types, numeric ranges, string lengths, and character classes. A new Profile Analysis view in Security Analytics lets teams review conformance trends and sampled violation details before enforcing, and profiles can then be enforced via Custom Rules scoped by hostname, path, operation, or other security signals such as Attack Score.

Review Schema Profile conformance trends and violation samples in Security Analytics before committing to enforcement.
📍In the Cloudflare dashboard, go to Security › Analytics › Profile Analysis to inspect conformance trends and sampled non-conforming requests for your Schema Profile.
— Names mechanism, sub-features, and exact dashboard path WAF - Enforce positive security with Applica…
03 00
Batch deletion of Workflow instances new 65

Enables deleting one or up to 100 Workflow instances and their stored state in a single operation via a binding or Wrangler.

— Names the limit and both entry points, no exact command product docs
04 00
Hyperdrive support for Python Workers new 65

Adds Hyperdrive support for Python Workers, enabling connection pooling and query caching for PostgreSQL and MySQL.

— Names supported databases and mechanism, no exact setup steps product docs
thinner coverage below
05 00
cloudflared drops 32-bit Windows and Intel macOS deprecated 55

Starting in 2027, cloudflared will stop publishing releases for 32-bit Windows and Intel-based macOS architectures.

— Gives exact timeline and affected architectures product docs
06 00
RPC session spans in Workers traces new 50

Automatically includes JavaScript RPC session spans in Workers traces for Worker-to-Worker and Worker-to-Durable Object calls.

— Automatic feature, no configuration step to act on product docs
07 00
Inspect panel for Browser Run Session Recordings new 45

Adds an Inspect panel to Browser Run Session Recordings, exposing console output, network activity, and page structure for post-session debugging.

— Names the panel but no navigation path or command product docs
08 00
Reject-busy mode for synchronous inference new 45

Adds a reject-busy mode for synchronous inference requests that fails immediately instead of waiting for capacity.

— Describes behavior but not how to enable it product docs
09 00
Unified Routing GA in Cloudflare WAN improved 30

Adds Unified Routing as generally available across supported WAN on-ramps in Cloudflare WAN.

— States GA status only, no mechanism or usage detail product docs
└──▷ BREAKING ON UPGRADE
  • !
    Starting in 2027, cloudflared will stop publishing releases for 32-bit Windows and Intel-based macOS architectures.
WAF - Enforce positive security with Application Profiles (2026-09-07) Blog Stable

Cloudflare WAFWAFA reverse proxy that inspects HTTP/HTTPS traffic and blocks requests matching known attack patterns such as SQL injection or XSS, sitting between clients and a web application to filter malicious input before it reaches the server. gains Application Profiles for positive-security enforcement, starting with Schema Profiles that learn and validate request structure.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

└──▷ HOW TO FIND IT
Review Schema Profile conformance trends and violation samples in Security Analytics before committing to enforcement.
📍In the Cloudflare dashboard, go to Security › Analytics › Profile Analysis to inspect conformance trends and sampled non-conforming requests for your Schema Profile.
  • Introduces Application Profiles, a new positive-security layer for Cloudflare WAF that learns expected request structure and classifies traffic as conforming or non-conforming without blocking.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
  • Adds Schema Profiles as the first profile type, modeling path variables, query parameters, headers, cookies, JSON bodies, and form-encoded bodies — including field types, numeric ranges, string lengths, and character classes.
  • Adds Profile Analysis view in Security Analytics to review conformance trends and sampled violation details before enforcing a profile.
  • Enables enforcement of Application Profiles via Custom Rules, allowing scope by hostname, path, operation, or other security signals such as Attack Score.

4 details in Everything

Was this useful?
// END OF TRANSMISSION

The Cyber Toolchain · curated daily · September 21, 2026
Releases & features from the open-source and commercial cyber tools that practitioners actually run.

▷ Subscribe
cyber-toolchain / feedback

Feedback on

compare.md

    Free the public facts, no sign-in Any paid plan all other sections Researcher & Business pricing history and category rank

    my-toolchain — 0 tools

      Betas, release candidates and previews. Turn it off to see final releases only. It also applies to changes you ask for over MCP when you're signed in.
      browse all tools → score card for your toolchain →
      cyber-toolchain / feedback

      Send us feedback