The Cyber Toolchain — issue 135, November 10, 2026
THE CYBER TOOLCHAIN
NO. 135
Tail
THE DAILY RELEASE FIREHOSE
PUBLISHED NOVEMBER 10, 2026 · EVERY DAY
EDITIONStail|grep|head|ls|diff|uniq
The daily firehose — everything the toolchain shipped today, already filtered.
We read every release from the 1325 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.
READ AS
Focus:What features shipped and what problems they solve, plus launches, packaging and where the market is moving.
Leaves out:Command-line flags, API-only deltas, usage examples, and releases that are only refinements.
Focus:New security controls, new APIs, and product improvements you can put to work, plus anything an upgrade breaks.
Leaves out:Launch and marketing news, governance, pricing, and documentation changes with no operational effect.
Focus:Dashboards, reporting, and anything you can quantify: metrics, coverage, risk scores, compliance impact, vendor moves.
Leaves out:Settings, flags, API and command-line detail, library bumps, and minor releases with no reporting or posture angle.
Focus:API, SDK and MCP changes, new libraries, breaking changes in what you import, and cyber tooling.
Leaves out:Governance, SOC operations, and vendor launches with no developer surface.
AMOUNT
DETAIL
ORDER
This issue came out before editions were ranked, so it shows the Standard edition in issue order.
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. Socket now builds SBOMsSBOMsA software bill of materials, an inventory of the components and versions inside an application. It helps teams find affected software when a dependency has a vulnerability or licensing issue. from Gradle, sbt, Maven and .NET build output instead of manifests, so fixes and reachability reflect what actually compiled. Lakera's second-stage model reads your system prompt before flagging a prompt attack, cutting false positives 95% on average. Yubico's Windows background service drops the admin elevation FIDO2FIDO2The open standard behind passkeys and hardware security keys, where a website checks a signature from the user's device instead of a password. Because the key is bound to the site, phishing pages cannot use it. used to demand.
identify
Get a dependency inventory from what the build actually produced
Socket infers SBOMs per build root from Gradle, sbt, Maven and .NET build output, closing the long-standing gap where a manifest lists a version range and the resolved graph contains something else. ZeroPath pairs reachability-aware scanning across 13 ecosystems with AI-BOMAI-BOMAn inventory of every AI model, dataset, and dependency used in a system, analogous to a software bill of materials, giving cyber tools a structured way to audit AI supply-chain risk. and container image inventory, so the list of what you run includes the model and image layers, not just the lockfile.
Stop babysitting prompt-attack alerts and FIDO2 elevation prompts
Lakera's two-stage Prompt Defense passes borderline input to a larger model with the application's own system prompt as context, which is the only way to tell a jailbreak from a user who legitimately asks the app to ignore instructions — 95% fewer false positives on average means the detector can stay in blocking mode. Yubico Authenticator 7.5.0's Windows background service removes the admin elevation FIDO2 operations required, which was the practical reason key management stalled on locked-down laptops; PIVPIVA US federal smart-card standard (FIPS 201) that stores cryptographic keys and certificates on a physical card, letting cyber tools authenticate users or sign data without exposing private keys. also gains ML-DSAML-DSANIST's main post-quantum digital signature standard (FIPS 204), formerly Dilithium. It is the planned replacement for RSA and ECDSA signatures on certificates and code./ML-KEMML-KEMNIST's post-quantum key-exchange standard (FIPS 203), formerly Kyber. Two parties use it to agree on a shared secret, usually alongside X25519 in a hybrid TLS handshake..
Issue short-lived identity to agents and sessions instead of handing out keys
JumpCloud's Agent Delegated Access mints signed JWTsJWTsJSON Web Tokens — an open standard (RFC 7519) for encoding claims as a signed, optionally encrypted JSON object. Cyber tools use them for stateless authentication without server-side session storage. so an AI agent acts as a scoped identity you can revoke, rather than inheriting a human's long-lived token. P0 Security binds OAuth, native and impersonation credentials to a client owner and session, so a leaked credential is useless outside the session that minted it, and 1Password Environments gives a project a shareable secret collection reachable from CLI, SDK and 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 instead of a shared vault.
Three unrelated things stand out today: ZeroPath added reachability-aware dependency scanning so SCASCASoftware composition analysis, identifying third-party dependencies and checking them for known vulnerabilities and license issues. It covers software a team uses rather than writes itself. noise can be tied to exploitable code paths; safe-chain can now scan internal-only apps through a self-hosted broker; Lakera added contextual second-stage prompt-attack review to cut false positives.
protect
Prioritize vulnerable dependencies by whether the code is actually reachable
ZeroPath added reachability-aware dependency scanning across 13 ecosystems, while Socket can infer SBOMsSBOMsA software bill of materials, an inventory of the components and versions inside an application. It helps teams find affected software when a dependency has a vulnerability or licensing issue. from build output instead of only manifest files. Teams get a better fix queue: fewer theoretical package findings, more emphasis on dependencies that are present in the built artifact and connected to reachable paths.
Scan applications and repositories that were previously awkward to reach
safe-chain added a self-hosted Broker for internal-only applications, and skylos added CI setup and branch-gating workflows around local scans. That moves checks closer to private apps and pre-merge code, instead of waiting for externally reachable targets or post-merge review.
Reduce prompt-attack alert fatigue with contextual review
Lakera added a second detection stage that uses a larger contextual model and the application’s system prompt. The practical gain is fewer benign prompts getting treated as attacks, so teams can keep prompt defenses tighter without drowning reviewers in false positives.
Stop treating AI gateway credentials as reusable shared secrets
P0 Security now binds OAuth, native, and impersonation credentials to client owners and sessions, with replay-protection controls. That narrows credential reuse paths in AI gateway flows, where one leaked or replayed token can otherwise cross user, client, or session boundaries.
Hacktron adds a full pentest and PR-review API, 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. tool suite, scheduled whitebox scans, bidirectional Jira sync, attestation letters, and finding IDs across all source control platforms.
Run a pentest end-to-end through the API with POST /pentests, PATCH /pentests/{id}, POST /pentests/{id}/login-checks, POST /pentests/{id}/estimate, and POST /pentests/{id}/start.
List open pull requests and trigger a Hacktron review programmatically with GET /pull-requests, GET /pull-requests/details, and POST /pull-requests/review.
Read and change PR review settings, including scm_check_statuses_enabled, and toggle reviews per repository with GET /pr-review-settings, PATCH /pr-review-settings, and POST /repos/pr-reviews.
Stop a scan, upload a code archive, discard a draft, and fetch credit and finding totals with POST /scans/{id}/stop, POST /archives/uploads, DELETE /pentests/{id}, GET /credits, and GET /findings/counts.
Start a fix pull request for a finding, track its progress, and check automation credits with POST /findings/{id}/remediate, GET /findings/{id}/remediation, and GET /credits/automation.
+21 moreshow less
Fetch pentest reports, open code checkouts, and manage applications through new endpoints including GET /scans/{id}/report, POST /code-workspaces, and POST /applications.
Draft, price, start, stop, and discard pentests through MCP tools create_pentest_draft, update_pentest_draft, estimate_pentest, start_pentest, stop_scan, discard_pentest_draft, and request_archive_upload.
Verify that Hacktron can sign in to each pentest site using the new verify_pentest_login MCP tool; pentest drafts now accept login targets for signed-in testing.
Let an AI assistant see repositories, cost estimates, credit balance, and finding totals through new MCP tools list_repositories, get_cost_estimation, get_credit_balance, and count_findings.
Start and configure PR reviews from an AI assistant using MCP tools list_pull_requests, get_pull_request, start_pr_review, get_pr_review_settings, update_pr_review_settings, and set_pr_reviews.
Generate and monitor fix pull requests and check automation credits from an AI assistant with MCP tools start_remediation, get_remediation, and get_automation_credit_balance.
Let agents read code and pentest reports and edit threat models and applications through MCP tools list_context_documents, read_context_document, get_pentest_report, open_code_workspace, edit_context_document, update_threat_model, and create_application.
API finding endpoints now accept a finding's short ID such as ACME-FD-12 in place of its UUIDUUIDA universally unique identifier, usually a 128-bit value used to identify an object across systems. It is designed to make independently generated identifiers unlikely to collide., and return it as display_id.
MCP finding tools now accept a finding's short ID such as ACME-FD-12 in place of its UUID.
Schedule whitebox scans daily, weekly, or monthly with the new On a schedule automation trigger, reusing saved scope, login instructions, and a credit limit from the Whitebox scan wizard.
New integrationTurn off Hacktron check statuses on GitHub, GitLab, Bitbucket, and Azure DevOps independently of review comments with the new Update check statuses in source control setting.
Updated integrationControl per-direction Jira sync with new Update tickets from findings and Update findings from tickets settings.
Findings with an open fix PR now show a Fix in review badge, and their activity lists pull requests, commits, and Slack messages that mention the finding's ID.
Threat models now refresh daily at 03:00 UTC when the default branch changes, and a single false-positive or accepted-risk decision triggers an update.
Whitebox retests now start from Request retest on a completed scan; Download report then offers Original scan and After retest editions.
A finding's OWASP Top 10:2025OWASP Top 10:2025A ranked list of the ten most critical web application risk categories, maintained by the Open Worldwide Application Security Project, used by cyber tools to prioritize detection, testing, and remediation coverage. category can now be set or cleared from its menu.
New attestation letters are available with a shareable link that opens without signing in and a downloadable PDF.
New Post system comments setting posts notices about skipped reviews, usage limits, and configuration problems.
Organisation admins can now edit, verify, estimate, start, and stop their teammates' pentests.
List repositories API now returns openFindingsCount and lastScanAt for each repository.
With dynamic testing, Define target now accepts several apps, each with its own domain, Auth instructions, and Verify access.
The official changelog for the Snyk AI Trust Platform. Discover what's new in our security intelligence, AI-powered tools, and latest product enhancements.
Snyk shipped a Customer Gateway for scanning internal applications behind the network perimeter, alongside risk-based 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. Governance policies, a new MCP Risk Score, finer tenant role scoping, and signed installers for the ADSADSAlternate Data Streams: an NTFS feature that lets a file carry hidden secondary data blobs under the same path, giving cyber tools a place to stash metadata or detect covert storage by malware. agent.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
6 FEATURES0 OF 6 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Signed ADS installers for macOS and Windowsnew90how completely this was documenteddepth30/40specificity30/30actionability30/3090 / 100
The ADS installer now ships as a signed macOS .pkg and Windows .msi, downloadable from Settings in Evo, and installable via sudo installer -pkg <PACKAGE_FILE> -target / or Start-Process msiexec.exe with TENANTID and PUSHKEY arguments.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Install the ADS agent on a macOS machine using the signed package, supplying your tenant credentials so it registers automatically.
— Exact install commands and arguments make it directly runnable2026-11-10-29977eee
0200
Customer Gateway for internal application scanningnew75how completely this was documenteddepth30/40specificity25/30actionability20/3075 / 100
Create a Customer Gateway from the CLI or from Settings and run it in your environment with Docker, Kubernetes, or the CLI to let Snyk Continuous Offensive Security reach applications only accessible inside your network. Greybox assessment is available through the gateway when repositories are linked, and every authentication method works through the gateway at the same depth as public targets.
— Names deployment methods and auth parity but no exact command2026-11-10-29977eee
0300
Risk-based policies in MCP Governancenew70how completely this was documenteddepth30/40specificity25/30actionability15/3070 / 100
MCP Governance now supports risk-based policies, letting you classify servers with an allowlist, define conditions on Evo risk dimensions such as 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. in a tool or unofficial publisher status, and preview blocking impact before enabling it.
— Describes mechanism but only via UI, no config surface named2026-11-10-29977eee
0400
MCP Risk Score in Agent Supply Chain Securitynew60how completely this was documenteddepth25/40specificity20/30actionability15/3060 / 100
Agent Supply Chain Security now surfaces an MCP Risk Score combining server behavior, trust, and local-server implementation signals, usable as a policy condition alongside any individual risk index.
— Explains composition but not scoring mechanics or usage steps2026-11-10-29977eee
thinner coverage below
0500
Enable and disable Evo policiesnew55how completely this was documenteddepth25/40specificity15/30actionability15/3055 / 100
Evo by Snyk now lets you disable and enable any policy, including default policies; a disabled policy stops raising new issues and hides previously raised issues until re-enabled.
Organization-scoped Tenant roles for Evoimproved55how completely this was documenteddepth20/40specificity20/30actionability15/3055 / 100
The Tenant Member role now has read-only Evo access scoped to the user's Organizations, and both Tenant Member and Tenant Viewer roles can be scoped by Organization.
GitHub is a web-based platform for version control and collaborative software development using Git repositories.
GitHub added Claude Fable 5.2 model support to Copilot across its surfaces and extended secret scanningsecret scanningChecking files, code or other data for exposed credentials such as passwords, API keys and tokens. Findings help teams remove the secret and revoke or rotate it. to catch leaked secrets in GitHub Actions workflow logs.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
2 FEATURES1 OF 2 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0101
Secret scanning for Actions workflow logsnew85how completely this was documenteddepth32/40specificity25/30actionability28/3085 / 100
Secret scanning now scans GitHub Actions workflow logs for provider patterns when a workflow run completes, in addition to issues, wikis, discussions, and pull requests.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Findings appear in the repository's secret scanning alert list alongside code alerts, with a link back to the specific workflow log; validity checks run for supported secret types, and alerts remain available even after the workflow logs are deleted. Enable it via enterprise settings under Policies > Advanced Security, setting the All repositories dropdown to Enabled.
Enable Actions workflow log scanning for all repositories across your enterprise to catch secrets leaked into CI output.
📍In the enterprise settings, go to Policies › Advanced Security, then set the All repositories dropdown to Enabled under Secret scanning for Actions workflow logs.
Claude Fable 5.2 in Copilotnew83how completely this was documenteddepth30/40specificity27/30actionability26/3083 / 100
GitHub Copilot now supports Anthropic's Claude Fable 5.2 model for complex, multistep coding tasks across large codebases, available in Visual Studio Code, Visual Studio, Copilot CLI, GitHub Copilot coding agent, the Copilot app, github.com, GitHub Mobile (iOS and Android), JetBrains IDEs, Xcode, and Eclipse via the model picker. Copilot Enterprise and Business admins must enable the Claude Fable 5.2 policy in Copilot settings (off by default); eligible enterprise customers can get zero data retention via a time-bound exemption while Anthropic rolls out Enterprise Frontier Safeguards (EFS). It's available to Copilot Pro+, Max, Business, and Enterprise users, billed at provider list pricing under usage-based billing.
Enable Claude Fable 5.2 for your organization so members can select it in the model picker across supported surfaces.
📍In GitHub Copilot settings (Enterprise or Business admin), navigate to Policies, find the Claude Fable 5.2 policy, and toggle it on. Note: enabling constitutes acknowledgment of Anthropic's data retention requirement.
Fcaptcha is a self-hosted, invisible CAPTCHA that detects AI agents driving real browsers. 2 AA, and EAA: no cookies, no third-party data sharing, and no visual or audio puzzle.
fcaptcha's latest release adds two research-oriented capabilities: bounded stdin imports for ephemeral exports and a research-only replay mode with per-sample isolation.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
2 FEATURES0 OF 2 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Research-only replay mode with per-sample isolationnew39how completely this was documenteddepth22/40specificity12/30actionability5/3039 / 100
fcaptcha adds a research-only replay mode with per-sample isolation and no gates or persistent JSON, intended for research use rather than production.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
— Names mechanism but no command or config surface givenv1.47.3
0200
Bounded stdin imports for research exportsnew23how completely this was documenteddepth15/40specificity5/30actionability3/3023 / 100
fcaptcha now accepts bounded stdin imports for ephemeral research exports.
— Describes capability only in prose, no flag or limit namedv1.47.3
DataDome protects web, mobile, and API applications by detecting and blocking bot and fraud traffic.
DataDome's Google Cloud Load Balancer module can now read the request host from the X-Forwarded-Host header instead of the default host.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
1 FEATURE0 OF 1 FEATUREmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
X-Forwarded-Host support in GCLB modulenew75how completely this was documenteddepth20/40specificity25/30actionability30/3075 / 100
The Google Cloud Load Balancer module can read the request host from the X-Forwarded-Host header instead of the default host by setting DATADOME_USE_X_FORWARDED_HOST in the module configuration.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ This is useful when the load balancer sits behind a reverse proxy that sets X-Forwarded-Host, so DataDome evaluates the correct original hostname.
When your Google Cloud Load Balancer sits behind a reverse proxy that sets X-Forwarded-Host, enable this setting so DataDome evaluates the correct original hostname rather than the proxy's host.
Infisical is the open-source platform for securing secrets, certificates, access, and AI Agents.
Infisical's v0.167.7 release adds native honey token management alongside new integrations for ClickHouse, OVHcloud, and Databricks, plus PKIPKIA framework of policies, roles, and cryptographic standards for issuing and managing digital certificates that bind public keys to identities, letting tools verify authenticity and encrypt communications without pre-shared secrets. and Agent Vault improvements.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
7 FEATURES1 OF 7 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
ClickHouse rotation, OVHcloud auth, Databricks gateway supportnew65how completely this was documenteddepth20/40specificity30/30actionability15/3065 / 100
Adds ClickHouse credentials rotation to the secret rotation engine, token authentication for OVHcloud KMS connections, and gateway routing support for Databricks app connections and syncs.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
— Three named integrations but no setup steps givenv0.167.7
0200
Agent Vault multi-project and access managementimproved65how completely this was documenteddepth25/40specificity25/30actionability15/3065 / 100
Agent Vault now supports multiple projects per organization, exposes inline access management in the Manage Access sheet, shows inline services on sessions, and lets each Agent Vault member sign in independently to MCP OAuthMCP OAuthAn authentication layer for the Model Context Protocol that lets MCP clients obtain delegated access tokens so cyber tools can call MCP servers on a user's behalf without handling raw credentials. services.
— Names UI surface but no API or config detailv0.167.7
thinner coverage below
0301
Native honey token managementnew50how completely this was documenteddepth25/40specificity15/30actionability10/3050 / 100
Infisical now manages honey tokens natively, letting teams deploy and revoke AWS IAMAWS IAMAWS Identity and Access Management is Amazon's service for controlling who and what can call AWS APIs, by attaching policies that allow or deny actions on resources. Cyber tools use it to scope credentials to least privilege.-based canary credentials without external tooling.
— Describes mechanism but no UI path or command givenv0.167.7
0400
Cross-signed internal CA certificatesnew40how completely this was documenteddepth20/40specificity10/30actionability10/3040 / 100
Support cross-signed certificates for internal CAs, enabling trust chain flexibility across PKI hierarchies.
— Describes benefit but no config or command shownv0.167.7
0500
DigiCert certificate reissue on same orderimproved40how completely this was documenteddepth20/40specificity10/30actionability10/3040 / 100
Reissue DigiCert TLS certificates on the same order, avoiding new order creation on renewal.
— Clear behavior change but no command shownv0.167.7
0600
Collapsible recently deleted projects listimproved40how completely this was documenteddepth15/40specificity10/30actionability15/3040 / 100
Recently deleted projects now appear as a collapsible group in the project list.
— Clear UI location but minimal functional detailv0.167.7
0700
Zstd compression for secret cacheimproved35how completely this was documenteddepth15/40specificity15/30actionability5/3035 / 100
Compress secret cache payloads with zstd to reduce cache size.
— States mechanism but no measurable impact givenv0.167.7
OWASP Juice Shop: Probably the most modern and sophisticated insecure web application
Juice Shop v20.4.0 introduces 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./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. passwordless authentication alongside two new passkey-themed hacking challenges, and overhauls Score Board filtering with technology tags and grouped categories.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
4 FEATURES0 OF 4 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Technology tags and grouped Score Board filtersimproved68how completely this was documenteddepth25/40specificity28/30actionability15/3068 / 100
Tags challenges with new Technology filter tags including SQL, NoSQL/MongoDB, JWTJWTAn open standard (RFC 7519) for encoding claims as a signed, optionally encrypted JSON object passed between parties. Cyber tools use it to verify identity and permissions without querying a central session store., OAuth, 2FA/TOTP, and XML, and organizes all Score Board filter tags into named groups — Constraints, Flavor, Learning, Technique, and Technology — with translatable labels and delayed tooltips.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Tutorial and Coding Challenge buttons on challenge cards are now highlighted when filtering by their corresponding tag.
— Names tags and groups, UI path implied but not exactv20.4.0
0200
Passkey/WebAuthn passwordless authenticationnew60how completely this was documenteddepth25/40specificity20/30actionability15/3060 / 100
Adds Passkey/WebAuthn passwordless authentication with a dedicated passkey management screen that displays the provider name and icon, plus a new passkey-user customer account for use with passkey challenges.
— Names screen and account but no setup stepsv20.4.0
thinner coverage below
0300
Passkey Bypass and Passkey Hijack challengesnew30how completely this was documenteddepth10/40specificity15/30actionability5/3030 / 100
Adds a Passkey Bypass five-star challenge and a Passkey Hijack five-star challenge, each with an accompanying coding challenge.
— Named challenges but no mechanism describedv20.4.0
0400
Edit button for AI chat messagesnew25how completely this was documenteddepth10/40specificity5/30actionability10/3025 / 100
Adds a button to edit user messages in the AI chat.
Dalfox is an XSS scanner that automates parameter analysis, reflected/stored/DOM-based XSS detection, and WAF fingerprinting for web apps.
Dalfox adds 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./Postman spec scanning, actionable confidence filtering with --min-confidence, blind OOBOOBOut-of-band communication: a separate channel from the primary data path, used to send control signals or exfiltrate data without touching monitored traffic, giving cyber tools a covert or resilient side channel. and session-loss parity on server/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., per-parameter filter fingerprints in all output formats, repeatable --sxss-url, and SCAN_TIMEOUT error codes.
└──▷ GET THIS VERSIONHomebrewDockergit
Homebrew
$ brew install dalfox
# already installed? upgrade to this version:$ brew upgrade dalfox
Docker
$ docker pull ghcr.io/hahwul/dalfox:v3.3.0
git
$ git clone --branch v3.3.0 https://github.com/hahwul/dalfox.git
# already have the repo? check out this version:$ git checkout v3.3.0
Homebrew lists this version · published under the project's GitHub account · checked 10 Nov
New integrationScan OpenAPI 3.x / Swagger 2.0 specs or Postman collections with -i openapi or -i postman, and retarget the server URL with --base-url.
Filter findings by confidence across CLI, REST, MCP, --limit, and exit code with --min-confidence <low|high>; meta.min_confidence.dropped prevents filtered targets from appearing clean.
Server and MCP now support blind OOB (blind_oob, blind_oob_wait) and session-loss detection (session_check, session_check_url), matching CLI behavior.
Findings now include a per-parameter filter fingerprint (allowed, blocked, escaped characters) in every output format.
--sxss-url is now repeatable (config key sxss_url takes a list); file-upload injections now target both file content and filename.
+2 moreshow less
Targets cut by --scan-timeout report a SCAN_TIMEOUT error code in target_summary and REST/MCP error_message.
Per-target state is keyed by URL + method + body; target_summary and meta.skipped gain a method field.
CAPEv2 is a malware Configuration And Payload Extraction
CAPEv2 adds a max_executables cap for archive analysis packages and configurable static preview with early dynamic analysis stop in the web UI︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ USE IT
Cap how many executables CAPE extracts and runs from a submitted archive to bound analysis time and resource use.
ini
max_executables=5
Prevent duplicate auto-mode processor supervisors from racing on the same completed-task queue — the lock file is created automatically; verify it exists after starting the processor in auto mode.
$ python utils/process.py auto
Limit the number of executables processed from an archive submission with the new max_executables option in the archive analysis package.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Configure static preview behaviour and enable early dynamic analysis stop from the web UI via the new configurable static preview feature (feat(web): configurable static preview and early dynamic analysis stop).
CAPEv2 now bundles data/7zz (7-Zip 26.06 x86_64, mode 0755) directly in the core repository so modules such as file_extra_info, msi_extract, suricata, and web zip downloads work without fetching from community.
CAPEv2 now preloads libmagic handles once at module import time in lib/cuckoo/common/path_utils.py via _preload_magic(), and adds shared magic_from_buffer() and magic_from_file() helpers with an os.memfd_create fallback for ZIP and other EOF-indexed formats misidentified by file 5.46+.
Add pcodedmp integration to Office static analysis: when pcodedmp is installed, the parse_office module now uses it for P-code decompilation alongside the existing oletools stack.
+1 moreshow less
Temp files in cleaners, distributed tasks, and GCP workers are now routed to cape-external within the directory set by [cuckoo] tmppath, and free_space_monitor is hardened to scan only the subdirectories cuckoo-tmp, cape-external, cuckoo-sflock, and cape-pubsub.
Limit the number of executables processed from an archive submission with the new max_executables option in the archive analysis package.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Configure static preview behaviour and enable early dynamic analysis stop from the web UI via the new configurable static preview feature (feat(web): configurable static preview and early dynamic analysis stop).
Add pcodedmp integration to Office static analysis: when pcodedmp is installed, the parse_office module now uses it for P-code decompilation alongside the existing oletools stack.
Finite State Platform is a firmware and software supply chain security tool that analyzes binaries and embedded devices to identify vulnerabilities, SBOMs, and risks.
Finite State Platform now documents filter and sort fields on list endpoints, returns stable 400 error codes for invalid filters, and flags estimated Audit Trail totals via a response header.
Filter and sort fields, operators, and forms on list endpoints are now documented; an invalid filter or sort returns HTTP 400 with the valid fields and a stable code — FILTER_FIELD_UNKNOWN, FILTER_OPERATOR_INVALID, FILTER_INVALID, SORT_FIELD_UNKNOWN, or SORT_INVALID — so callers can match on the code when message text differs.
On GET /public/v0/audit, GET /public/v0/audit-events, and GET /public/v0/projects/{projectId}/audit, a total above 10,000 is flagged as an estimate via the X-Total-Count-Estimated: true response header, and the Audit Trail now shows totals above 10,000 as 10,000+.
Netcap is a framework for secure and scalable network traffic analysis - https://netcap.io
Netcap's latest releases add a full net firewallWAFWAFA 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. command with policy-driven audit responses, new net split and net mcp commands, TLS 1.2/1.3 decryption via embedded NSSNSSNetwork Security Services, a set of cryptographic libraries maintained by Mozilla that implement TLS, PKI, and related standards; cyber tools use it to get a battle-tested, FIPS-validated crypto layer without writing their own. key logs, and a string of capture-reliability, detection, and database-management improvements.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
17 FEATURES3 OF 17 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Audit-record response policies in net capturenew83how completely this was documenteddepth30/40specificity28/30actionability25/3083 / 100
Apply explicit audit-record response policies during capture with --firewall-policy and dispatch live response effects over -iface captures with --firewall-effects-config on net capture.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Print the versioned audit policy field types, sensitivity, and projection bounds as JSON using --firewall-schema.
Inspect the firewall's audit policy schema before writing a policy file, to see which fields, sensitivity levels, and projection bounds are available.
$ net capture --firewall-schema
Apply a response policy to a completed capture and dispatch live effects on an active interface capture simultaneously.
$ net capture -iface eth0 --firewall-policy policy.json --firewall-effects-config effects.json
— Three flags named with runnable examples for each.v0.16.1
0201
New `net firewall` WAF commandnew78how completely this was documenteddepth35/40specificity28/30actionability15/3078 / 100
New net firewall command adds a WAF with routed HTTP/HTTPSHTTPSHTTP carried over an encrypted TLS connection. It protects traffic in transit and lets a browser verify the server's certificate. and API inspection, scoped identities and approvals, a Coraza/HTTP3 build profile, SMBSMBA network file-sharing protocol developed by Microsoft that lets systems share files, printers, and named pipes over a network; cyber tools target or speak it because it is ubiquitous in Windows environments./FTPFTPA decades-old TCP-based protocol for transferring files between a client and server, using separate control and data connections; cyber tools target or support it because it often transmits credentials in plaintext. protocol gateways, typed response actions, transactional control, an embedded policy management console, and Helm deployment examples.
— Lists many capabilities but no example invocation shown.v0.16.1
0300
Higher default stream reassembly limitimproved75how completely this was documenteddepth25/40specificity25/30actionability25/3075 / 100
The max-stream-bytes flag for net capture now defaults to 100 MiB (raised from 10 MiB), controlled via NC_MAX_STREAM_BYTES.
Cap stream reassembly memory in a 4 GiB container without overriding the cgroup-aware GC headroom Netcap now applies automatically.
— Exact flag, env var, default change, and runnable example.v0.16.3
0402
TLS 1.2/1.3 decryption via embedded NSS key logsnew70how completely this was documenteddepth30/40specificity25/30actionability15/3070 / 100
Netcap now decodes authenticated TLS 1.2 and TLS 1.3 HTTP streams using NSS key logs embedded in pcapng files, and marks decrypted records explicitly.
— Names NSS key log mechanism but no command shown.v0.16.3
0500
Signed database and enrichment management in net utilnew65how completely this was documenteddepth20/40specificity25/30actionability20/3065 / 100
Install signed enrichment databases with net util -download-dbs, refresh them with net util -update-dbs, and fetch optional enrichment data with net util --download-enrichment.
— Three exact subcommands named for db management.v0.16.1
0600
Gopacket fork adoption and startup reportingimproved60how completely this was documenteddepth25/40specificity25/30actionability10/3060 / 100
Upstream gopacket is replaced by the dreadl0ck/gopacket fork at commit f2eeabc1ff6e, adding live-only close wakeups and bounded readers, and Netcap now reports the effective gopacket replacement at startup.
— Names fork and commit but no migration steps.v0.16.3v0.16.1
0700
Content-based capture file identificationimproved60how completely this was documenteddepth25/40specificity20/30actionability15/3060 / 100
Netcap now identifies capture files by content (magic bytes) rather than .pcap/.pcapng extension, so inputs with non-standard extensions are accepted by net capture -read.
— Explains mechanism and flag but no limits given.v0.16.3
0800
Cgroup-aware memory limits via GOMEMLIMITimproved60how completely this was documenteddepth25/40specificity20/30actionability15/3060 / 100
Netcap now respects GOMEMLIMIT and applies cgroup-aware Go GC headroom, keeping capture within container memory limits.
— Names env var and behavior, no numeric limits.v0.16.3
thinner coverage below
0900
New `net split` capture subsetting commandnew55how completely this was documenteddepth25/40specificity20/30actionability10/3055 / 100
New net split command reads PCAPPCAPA packet capture, recorded network packets that can be inspected later. Analysts use it to reconstruct communication and investigate protocol behavior or suspicious traffic./PCAPNG captures and exports bounded packet subsets and time shards, including time-only exports without a native BPFBPFA Linux kernel virtual machine that lets programs attach small, sandboxed bytecode filters to kernel hooks — syscalls, network events, tracepoints — so cyber tools can observe or redirect kernel behaviour without writing a kernel module. runtime.
— Describes capability, no flag names or example given.v0.16.1
1000
Live-stream maintenance and flush intervalsnew45how completely this was documenteddepth15/40specificity20/30actionability10/3045 / 100
Control independent live-stream maintenance and output snapshot intervals with --live-maintenance-interval and --live-flush-interval on net capture.
— Names two flags, no defaults or example given.v0.16.1
1100
Checksum-pinned resumable corpus replaynew45how completely this was documenteddepth20/40specificity15/30actionability10/3045 / 100
Netcap adds checksum-pinned resumable corpus replay, allowing interrupted replay runs to continue from a known-good checkpoint.
— Mechanism named, no flag or command given.v0.16.3
1200
SMB DCE/RPC and Netlogon detection coverageimproved45how completely this was documenteddepth20/40specificity20/30actionability5/3045 / 100
Netcap now recovers additional SMB-carried DCE/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. operations and covers additional service-control and Netlogon operations in detections.
— Names protocols covered, no flags or examples.v0.16.3
1300
Beacon incident aggregation in c2.beaconimproved40how completely this was documenteddepth20/40specificity15/30actionability5/3040 / 100
Netcap now aggregates beacon incidents and detects skipped POST polling, extending c2.beacon coverage.
— Names c2.beacon area, no thresholds or flags.v0.16.3
1403
New `net mcp` MCP server commandnew35how completely this was documenteddepth15/40specificity15/30actionability5/3035 / 100
New net mcp command exposes a stdio 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 AI-agent workflows.
— One-line description, no protocol details or example.v0.16.1
1500
NBNS and IEEE 802.15.4 audit recordsnew30how completely this was documenteddepth10/40specificity15/30actionability5/3030 / 100
Netcap adds NBNSNBNSNetBIOS Name Service, a legacy Windows broadcast protocol on UDP port 137 that resolves hostnames to IP addresses on local networks; cyber tools exploit or monitor it to discover hosts and capture credentials. and IEEE 802.15.4IEEE 802.15.4A wireless communication standard defining the physical and MAC layers for low-rate, short-range personal area networks, used as the foundation for protocols like Zigbee and Thread in IoT and embedded device research. audit records, expanding the set of protocol-layer records produced during capture.
— Just names two new record types, no field detail.v0.16.3
1600
SLL2 RARP link-layer frame supportnew30how completely this was documenteddepth10/40specificity15/30actionability5/3030 / 100
Netcap now handles SLL2 RARP link-layer frames during packet decoding.
Plerion is a cloud security platform that detects misconfigurations, vulnerabilities, and threats across cloud environments, helping teams identify and remediate security risks.
Plerion significantly expanded its AWS inventory coverage, adding collection support for ten service families spanning OpenSearchOpenSearchAn open-source search and analytics engine, forked from Elasticsearch by AWS in 2021, that lets cyber tools store, query, and visualize large volumes of log or event data without licensing restrictions. Serverless, EKSEKSAmazon's managed Kubernetes service, run by AWS, that handles control-plane provisioning so teams can deploy containerized workloads without operating the underlying cluster infrastructure themselves., Route 53 Resolver, Global Accelerator, EventBridgeEventBridgeAWS's managed event bus service that routes events between cloud services, SaaS apps, and custom code using rules and targets, letting tools react to state changes without polling or managing message infrastructure., Aurora DSQL, MSK, Athena, Backup, and IoTIoTA collective label for internet-connected physical devices — sensors, cameras, routers, appliances — that tools target because they run embedded firmware with limited patching and broad network exposure. Core.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
10 FEATURES0 OF 10 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Aurora DSQL cluster collectionnew73how completely this was documenteddepth30/40specificity28/30actionability15/3073 / 100
Plerion now collects Amazon Aurora DSQL clusters, showing status, deletion protection, encryption type and 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, endpoint, multi-Region peers, and resource-based policy; requires dsql read permissions, present on stacks deployed or updated since July 2026.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
OpenSearch Serverless inventory collectionnew63how completely this was documenteddepth30/40specificity25/30actionability8/3063 / 100
Plerion now collects AWS OpenSearch Serverless collections, 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. endpoints, security configurations, and encryption, network, and data access policies — each showing its type, encryption key, standby replicas, deletion protection, endpoints, tags, and full policy or 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./IAMIAMIdentity and Access Management — a framework of policies and services for controlling who can authenticate to systems and what resources they are authorized to use, letting cyber tools enforce least-privilege and audit access. federation settings.
EventBridge connections, destinations, and rule attachmentnew61how completely this was documenteddepth28/40specificity25/30actionability8/3061 / 100
Plerion now collects Amazon EventBridge connections and API destinations as standalone assets, and attaches up to 300 rules per event bus — each rule showing its event pattern or schedule, state, IAM role, and targets.
MSK VPC connection collection and faster policy updatesimproved60how completely this was documenteddepth28/40specificity24/30actionability8/3060 / 100
Plerion now collects Amazon MSK managed VPC connections, showing the target cluster, VPC, subnets, security groups, authentication type, and state; MSK clusters now also update within minutes when a cluster policy is deleted.
Route 53 Resolver resource collectionnew59how completely this was documenteddepth28/40specificity25/30actionability6/3059 / 100
Plerion now collects seven Route 53 Resolver resource types: endpoints, resolver rules, query logging configs, DNS Firewall rule groups, domain lists, and the VPC associations of rules and rule groups — including embedded rules, IP addresses, and sharing policies.
Athena workgroup and data catalog collectionnew59how completely this was documenteddepth28/40specificity25/30actionability6/3059 / 100
Plerion now collects Amazon Athena workgroups and registered data catalogs, showing engine version, query result location and encryption, data scan limit, catalog type, parameters, and status.
IoT Core resource collectionnew55how completely this was documenteddepth25/40specificity24/30actionability6/3055 / 100
Plerion now collects six AWS IoT Core resource types: policies, custom authorizers, role aliases, topic rules, CA certificates, and domain configurations — each with its operational settings, status, and tags.
EKS add-ons and identity resource collectionnew53how completely this was documenteddepth25/40specificity22/30actionability6/3053 / 100
Plerion now collects four additional Amazon EKS resource types: add-ons, Fargate profiles, 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. configurations, and Pod Identity associations, each with role, status, 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., and tag details.
Global Accelerator inventory collectionnew53how completely this was documenteddepth25/40specificity22/30actionability6/3053 / 100
Plerion now collects AWS Global Accelerator accelerators and their listeners, showing status, static IPs, DNS name, flow-log state, protocol, port ranges, and client affinity.
Plerion adds an Access tab to AWS assets showing inbound/outbound access paths and policy layers (SCPsSCPsAWS Service Control Policies — organization-wide guardrails from Amazon that cap the maximum permissions any IAM entity in a member account can hold, letting cyber tools enforce least-privilege boundaries across an entire AWS org., RCPs, permission boundaries) as a graph or list.
Open any AWS asset and use the new Access tab to see what can access the asset, what it can access, and who can gain its permissions — viewable as a graph or a list.
Each arrow in the access graph names the permission level it allows (read, write, or administer); hover over an arrow to inspect the policy layers on the path, including identity policies, permission boundaries, SCPs, RCPs, and the resource policy, with the matching statements from the policy document.
Distroless is a language focused docker images, minus the operating system.
distroless adds a parallel Go-based sign_and_push runner to replace the generated bash script︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ TRY IT
Run the new parallel sign-and-push pipeline with a custom concurrency level to push, verify, and tag images faster in CI.
$ ./bazelisk run :sign_and_push --config=release----keyless $KEYLESS --jobs 32
Adds a Go-based sign_and_push binary that wraps crane and cosign and runs image push and offline cosign verification concurrently, controlled by --jobs (default 16), against a single initialized trusted_root.json.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Attaches image_aspect directly to sign_and_push targets and reads digests from 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.index.json at runtime, dropping the intermediate filegroup, oci_image_spdx, and stamp-sensitive .digest targets.
2 details in Everything
└──▷ BREAKING ON UPGRADE
!
The intermediate `filegroup`, `oci_image_spdx`, and stamp-sensitive `.digest` Bazel targets are removed; build configs that depended on them directly will break.Our best guess
Anchore's software supply chain solutions automate vulnerability scanning, strengthen container security, and support compliance with NIST, FedRAMP & more.
Anchore Enterprise introduces SBOMSBOMA software bill of materials, an inventory of the components and versions inside an application. It helps teams find affected software when a dependency has a vulnerability or licensing issue. drift detection spanning build and runtime: a new tag_drift policy gate catches package changes between builds, while the anchore-k8s-inventory Helm agent tracks what's actually running in Kubernetes clusters.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ It also adds SPDXSPDXAn open standard for describing software components, licenses and their relationships. Its documents let different tools exchange a software bill of materials without inventing a new format./CycloneDXCycloneDXAn open standard for describing software components and their dependencies, vulnerabilities and other supply-chain information. Tools use it to exchange bills of materials and related evidence. SBOM export and routes policy evaluation updates to Slack, Jira, GitHub, Teams, email, or webhook.
└──▷ WHAT SHIPPED ·
4 FEATURES0 OF 4 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Runtime SBOM drift via Kubernetes inventory agentnew95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
Deploy the anchore-k8s-inventory agent via Helm from the anchore/k8s-inventory chart at charts.anchore.io to poll the Kubernetes API on a configurable interval and report each container's image_digest, image_tag, pod_uid, and cluster name back to Anchore Enterprise.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Activating a runtime_inventory subscription triggers automatic vulnerability and policy analysis of every image the agent discovers, with findings surfaced in a Kubernetes-aware GUIGUIA graphical user interface, controls such as windows, buttons and menus. It lets people use software visually rather than entering commands. view broken down by cluster, namespace, node, and pod.
Deploy the runtime inventory agent into a cluster to get continuous visibility into what images are actually running, without requiring broad cluster credentials.
tag_drift policy gate for build-time driftnew90how completely this was documenteddepth30/40specificity30/30actionability30/3090 / 100
Anchore Enterprise adds a tag_drift policy gate that detects when a new build changes the packages present in an image compared to a prior build of the same tag. Scope the packages_modified trigger to OS-level packages only by setting the package_type parameter to apk, rpm, or deb, limiting noise from intentional language-dependency bumps; a policy rule can set an action such as STOP to block the build when drift is detected.
Block a build when any OS package changes between successive builds of the same image tag, using packages_modified scoped to apk packages.
Notifications for policy evaluation updatesimproved45how completely this was documenteddepth15/40specificity20/30actionability10/3045 / 100
Send policy evaluation updates — including tag_drift trigger results — to Slack, Jira, GitHub, Microsoft Teams, email, or a webhook via Anchore Enterprise Notifications.
SBOM export in SPDX and CycloneDXnew35how completely this was documenteddepth10/40specificity15/30actionability10/3035 / 100
Export SBOMsSBOMsA software bill of materials, an inventory of the components and versions inside an application. It helps teams find affected software when a dependency has a vulnerability or licensing issue. generated during image analysis in SPDX or CycloneDX formats.
Nudge Security is a SaaS security platform that discovers shadow IT assets, manages identity governance, and hardens security posture.
Nudge Security adds optional in-browser redaction enforcement so detected sensitive data can be blocked before it leaves the browser.
New detectionEnable one-click in-browser redaction of detected sensitive data, with optional enforcement so unredacted data never leaves the browser, via Settings > In-browser nudges.
Choose which regional data types to monitor by selecting active regions in Browser Extension Settings.
Sysdig Secure is a cloud-native security platform that detects threats, monitors runtime activity, and scans container images and Kubernetes environments for vulnerabilities and misconfigurations.
Sysdig Secure's Registry Scanner now supports credential-free authentication to Azure Container Registry on AKSAKSAzure Kubernetes Service, Microsoft's managed Kubernetes offering, which lets cyber tools offload cluster provisioning and scaling to Azure so engineers focus on workload logic instead of control-plane operations., along with clearer diagnostics for identity-related scan failures.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
2 FEATURES0 OF 2 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Managed/workload identity auth for ACR on AKSnew70how completely this was documenteddepth30/40specificity25/30actionability15/3070 / 100
Registry Scanner now authenticates to Azure Container Registry using managed identity or workload identityworkload identityAn identity assigned to software, such as a service or container, rather than a person. It lets the workload authenticate and receive permissions without sharing a user's credentials. on AKS, via the cluster's kubelet identity or a user-assigned identity federated to its service account, removing the need to store registry credentials.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Both RBACRBACRole-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. and RBAC + ABACABACAttribute-based access control, where rules evaluate facts about the user, resource and request. Access can depend on conditions such as department, data sensitivity or device status. repository permission modes are supported.
ACR identity authentication failure loggingimproved45how completely this was documenteddepth20/40specificity15/30actionability10/3045 / 100
Registry Scanner adds specific log messages for ACR identity authentication failures, distinguishing between an identity that cannot get a token, one with no role on the registry, and requests blocked by registry network rules.
Transform the enterprise SOC with Torq — combining agentic SecOps and Hyperautomation to triage, investigate, and respond to threats faster than ever.
Torq enhanced case management with Markdown-formatted case tabs and brought one-click quick actions directly to observables across multiple surfaces.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
2 FEATURES0 OF 2 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Quick actions on observablesnew60how completely this was documenteddepth25/40specificity20/30actionability15/3060 / 100
Quick actions now appear directly on observables — attach a workflow to any observable and it surfaces on the observable's own page, the case's Observables tab, and the hover card.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Markdown sections in custom case tabsnew45how completely this was documenteddepth20/40specificity15/30actionability10/3045 / 100
Custom case tabs now support free-form Markdown sections alongside custom fields, letting workflows populate each case tab with tables, formatted text, and structured reports.
Pysigma is a python library to parse and convert Sigma rules into queries (and whatever else you could imagine)
pySigma adds a new module for matching events against Sigma rules directly in Python, without needing to convert rules to a query language first.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
1 FEATURE0 OF 1 FEATUREmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Event matching via sigma.matching modulenew55how completely this was documenteddepth20/40specificity20/30actionability15/3055 / 100
The new sigma.matching module lets users match live or logged events directly against Sigma rules in Python code, without first converting those rules to a query language.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
— Names the module but no API details or usage examplev2.1.0
pySigma adds optional strict mode for field mapping transformations, preserves finalizer configs on pipeline load, and allows dashes in 1/all of selector patterns.
└──▷ GET THIS VERSIONpipgit
pip
$ pip install pysigma==2.0.3
# already installed? upgrade to this version:$ pip install --upgrade pysigma==2.0.3
git
$ git clone --branch v2.0.3 https://github.com/SigmaHQ/pySigma.git
# already have the repo? check out this version:$ git checkout v2.0.3
PyPI attestation · checked 10 Nov
Add an optional strict mode for field mapping transformations to catch unmapped fields at conversion time rather than silently passing them through.
Preserve finalizer configurations when loading processing pipelines, so pipeline-defined finalizers are no longer silently dropped on load.
Allow dashes in 1/all of selector patterns so rule conditions referencing hyphenated detection identifiers now parse correctly.
OpenCTI is an open Cyber Threat Intelligence Platform
OpenCTI's biggest addition this window is a platform-wide custom fields system that lets administrators extend entities and relationships with typed attributes usable across filters, CSV/JSON mappers, widgets, exports, and full-text search; alongside that it shipped Form Intake submission capabilities, configurable per-relationship deduplication windows, TLS/GraphQL security hardening, unified Filigran XTM Platform licensing, and smaller improvements to user management, dashboards, and date/time display.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
14 FEATURES5 OF 14 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0101
Custom fields for entities and relationshipsnew96how completely this was documenteddepth38/40specificity30/30actionability28/3096 / 100
Administrators manage custom fields (create, edit, delete, attach to entity types) in Settings > Taxonomies > Custom fields using the Manage custom fields capability, set per-entity-type defaults using @now for date fields, and mark fields mandatory (bypassed by Bypass mandatory attributes).︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Technical names start with x_opencti_cf_ (e.g. x_opencti_cf_priority) and are usable in the API, filters, exports, and STIXSTIXStructured Threat Information Expression, a standard format for describing threats, indicators and their relationships. It helps organizations exchange threat intelligence in a usable form. bundles; fields can be mapped at ingestion via aliases (STIX custom properties starting with x_), imported from STIX bundles, CSV mappers, and JSON mappers by technical name or alias, and exported in STIX exports/streams as properties in the object's OpenCTI extension. Custom fields work as filter keys in lists, dashboards, saved filters, streams, triggers, and playbooks with the same operators as standard attributes, as columns in list widgets when filtered to a single entity type, and in attribute widgets in Fintel templates and custom views; values can be bulk-updated via the Manipulate knowledge playbook component or the Update background task action. Seven value types are supported — Text, Markdown, Integer (with min/max bounds), Boolean, Date, Selection list, and Multiple selection list — attachable to Stix core relationship (covering every relationship type) or to Sighting; custom field values are also searchable via the platform's full-text search and a custom resolution field can be added to cases through this system. Without an Enterprise Edition license, existing values remain readable, filterable, importable, and exportable, but creating or managing fields requires the license.
Ingest a STIX bundle from a connector that uses a proprietary priority property, mapping it to a custom field via an alias so no connector-side changes are needed.
📍Define the custom field with technical name x_opencti_cf_priority and add x_mycompany_priority as an alias in Settings › Taxonomies › Custom fields. The bundle property x_mycompany_priority will then populate the custom field on ingest automatically.
Set a creation-date default on a custom date field so every new entity gets a timestamp without manual entry.
📍In Settings › Customization › Entity types, select the target entity type, go to its Attributes tab, attach the date custom field, then set its default value to @now using the edit icon.
— Names config path, capability, prefix, value types and config example7.261109.0
0202
TLS certificate verification controlsnew93how completely this was documenteddepth35/40specificity30/30actionability28/3093 / 100
New TLS verification settings govern outgoing connections platform-wide: tls:reject_unauthorized / TLS__REJECT_UNAUTHORIZED is a default for feeds, synchronizers, 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., RabbitMQ, HTTPSHTTPSHTTP carried over an encrypted TLS connection. It protects traffic in transit and lets a browser verify the server's certificate. proxy, and Filigran services (opt in early by setting it true before it becomes the default); https_proxy_reject_unauthorized / HTTPS_PROXY_REJECT_UNAUTHORIZED independently overrides the HTTPS proxy and otherwise inherits from tls:reject_unauthorized; and opencti:tls_reject_unauthorized / OPENCTI_TLS_REJECT_UNAUTHORIZED sets the default for workers and connectors, inherited by opencti:ssl_verify / OPENCTI_SSL_VERIFY and by worker:listen_api_ssl_verify for connectors in HTTP mode.
Opt in to strict TLS verification for all platform outgoing connections before it becomes the default, while also hardening GraphQL in one config block.
— Every config key and inheritance path named with a config exampleproduct docs
0303
Relationship deduplication window configurationbreaking88how completely this was documenteddepth32/40specificity28/30actionability28/3088 / 100
Configure the deduplication merge window globally or per relationship type using the relations_deduplication config key, with past_days and next_days accepting a number of days or an ISO 8601ISO 8601An international standard for representing dates and times as text, using formats like 2025-04-23T14:00:00Z. Tools adopt it to ensure timestamps are unambiguous and sortable across systems and locales. duration (e.g. PT30M, PT6H, P1DT12H), or override per type via environment variables such as RELATIONS_DEDUPLICATION__TYPES_OVERRIDES__STIX-SIGHTING-RELATIONSHIP__PAST_DAYS and RELATIONS_DEDUPLICATION__TYPES_OVERRIDES__STIX-SIGHTING-RELATIONSHIP__NEXT_DAYS set on every platform node. When an incoming entity carries no confidence level, the existing entity keeps its own confidence, or receives the sending user's maximum confidence if it has none. A per-type override replaces the global configuration entirely — any omitted past_days/next_days key resolves to 0, not the global default — and invalid duration values (e.g. 30m instead of PT30M) are silently ignored, also producing a window of 0 with no validation error.
Narrow the sighting deduplication window to 30 minutes to avoid merging sightings that are close in time but distinct events.
— Runnable config/env examples, but silent-failure behavior is the real riskproduct docs
0404
GraphQL security hardening defaultsnew88how completely this was documenteddepth32/40specificity30/30actionability26/3088 / 100
app:graphql:secure_by_default / APP__GRAPHQL__SECURE_BY_DEFAULT enables a secure posture that turns on GraphQL Armor, strips suggestions from error messages, disables the playground and introspection, and restricts WebSocketWebSocketA protocol that upgrades an HTTP connection into a persistent, full-duplex channel, letting a server push data to a client without repeated polling. Cyber tools use it for real-time event streaming and live command output. to subscriptions only. Alias-based batching protection caps are configurable via app:graphql:batching_protection:query_default / APP__GRAPHQL__BATCHING_PROTECTION__QUERY_DEFAULT, app:graphql:batching_protection:query_subtypes / APP__GRAPHQL__BATCHING_PROTECTION__QUERY_SUBTYPES, and app:graphql:batching_protection:mutation_default / APP__GRAPHQL__BATCHING_PROTECTION__MUTATION_DEFAULT, with the token mutation always hard-limited to 1.
Opt in to strict TLS verification for all platform outgoing connections before it becomes the default, while also hardening GraphQL in one config block.
— Names every config key and cap; example config block providedproduct docs
0500
Configurable date and time display formatsnew86how completely this was documenteddepth30/40specificity28/30actionability28/3086 / 100
Platform parameters now include Date format in tables (Default or ISOISOA widely adopted international standards body that publishes technical specifications; cyber tools cite ISO standards to signal compliance with recognized benchmarks for quality, security, or interoperability. 8601 YYYY-MM-DD) and Time in tables (Default, Date only, HH:mm, or HH:mm:ss), controlling how dates appear across every table and list platform-wide. Individual users can override this in their profile under User experience using the same two settings, with a Platform default option that defers to the platform setting.
Set the platform to display unambiguous ISO 8601 timestamps with seconds in every table — useful for a SOCSOCA security operations center, the team or function that monitors threats, investigates alerts and coordinates incident response. Its work can be internal, outsourced or shared. sharing incident lists with external customers.
📍In the platform administration, go to Parameters › (platform settings card) › set Date format in tables to ISO 8601 (YYYY-MM-DD) and Time in tables to Hours, minutes and seconds (HH:mm:ss). Dates will appear as e.g. 2026-10-27 15:32:17+01:00 across all entity and relationship lists, audit logs, alerts, and connector works.
Override the platform date format for your own account without affecting other users.
📍Click the top-right profile icon › Profile › User experience card › set Date format in tables and Time in tables to your preferred values. Choosing any value other than Platform default takes precedence over the platform setting.
— Exact UI navigation and setting values with sample output formatproduct docs
0600
Form Intake submission capabilitiesnew77how completely this was documenteddepth32/40specificity25/30actionability20/3077 / 100
A Create knowledge via Form Intake capability lets users submit Form Intakes to create knowledge without the general import-knowledge capability; submissions always create new records rather than updating existing ones, and only users who also hold Create / Update knowledge in the destination can update matched existing objects. A draft-scoped variant, 'Create knowledge via Form Intake (creation forced to draft)', forces all submissions into draft mode for restricted users. While filling in a form, users can create entities, observables, labels, and external references on the fly (with attached files and markdown images), held in the browser and created together with the rest of the submission only on submit; submissions are processed asynchronously, with a success message confirming the submission has been queued rather than that objects already exist.
Grant an analyst the ability to submit intelligence forms in draft only, so their intake goes into a review queue rather than directly into the main knowledge base.
📍In OpenCTI, go to Settings › Security › Users (or the relevant Role), then enable the capability Create knowledge via Form Intake scoped to draft mode. The user can now submit Form Intakes, and all submissions will land in a draft for review before promotion.
— Capability names and before/after given; UI path only sketchedproduct docs
0705
Elasticsearch 9 client selectionbreaking71how completely this was documenteddepth22/40specificity25/30actionability24/3071 / 100
Select the Elasticsearch/OpenSearchOpenSearchAn open-source search and analytics engine, forked from Elasticsearch by AWS in 2021, that lets cyber tools store, query, and visualize large volumes of log or event data without licensing restrictions. client explicitly with elasticsearch:engine_client / ELASTICSEARCH__ENGINE_CLIENT, accepting elasticsearch8, elasticsearch9, or auto-detection; this replaces the now-ignored engine_check configuration key, so deployments relying on engine_check must migrate.
Pin OpenCTI to the Elasticsearch 9 client when a proxy prevents the platform from reading the server version at startup.
yaml
elasticsearch:
engine_client: elasticsearch9
— Exact replacement key and migration example given, impact scope thinproduct docs
0800
Unified Filigran XTM Platform licensingnew69how completely this was documenteddepth25/40specificity22/30actionability22/3069 / 100
Paste a Filigran XTM Platform license PEM into the existing license field (settings page or APP__ENTERPRISE_EDITION_LICENSE) to activate Enterprise Edition across OpenCTI, OpenAEV, and OpenCRQ modules. Instances can also register with XTM One to activate Enterprise Edition without a local license; OpenCTI verifies the Filigran-signed license at every registration and shows Filigran license (through XTM One) as the license source on the settings page.
Activate Enterprise Edition on a self-hosted OpenCTI instance by supplying the Filigran XTM Platform license via environment variable instead of the UI.
— Env var and registration flow given, but activation internals unexplainedproduct docs
thinner coverage below
0900
Prometheus metrics endpoint host bindingnew48how completely this was documenteddepth12/40specificity18/30actionability18/3048 / 100
Bind the Prometheus metrics endpoint to a specific IPv4 address using app:telemetry:metrics:exporter_prometheus_host / APP__TELEMETRY__METRICS__EXPORTER_PROMETHEUS_HOST.
— Exact config key given but no example or use caseproduct docs
1000
Merge duplicate user accounts and lockout policynew42how completely this was documenteddepth18/40specificity14/30actionability10/3042 / 100
Administrators can merge duplicate user accounts — for example, merging a local account into an SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management.-created account — via the Merging users workflow in the user management interface, and can configure account lockout policy settings to control how repeated failed logins are handled.
— Names the workflow but gives no exact settings path or thresholds7.261109.0
1100
Time-series breakdown dimension in dashboard widgetsnew42how completely this was documenteddepth18/40specificity14/30actionability10/3042 / 100
The dashboard widget editor now supports configuring a time-series breakdown dimension, letting a time series be split by a chosen attribute in a single chart; this breakdown is supported in both private and public dashboards.
— Describes capability but no config path or example attribute7.261109.0
1200
Enrichment connector launch from investigation graphnew39how completely this was documenteddepth15/40specificity10/30actionability14/3039 / 100
Run an enrichment connector directly from the investigation graph's right-click menu without leaving the graph view.
— UI entry point named but no connector list or mechanism7.261109.0
1300
Ingestion health monitoring enhancementsimproved34how completely this was documenteddepth14/40specificity14/30actionability6/3034 / 100
Ingestion health monitoring now tracks last-run failures, feed and sync access-denied events, errors from ping, and disabled users.
— Lists tracked signals only, no UI path or alerting mechanism7.261109.0
1400
Label export/import in configuration presetsnew26how completely this was documenteddepth10/40specificity10/30actionability6/3026 / 100
Labels can now be exported and imported as part of configuration presets.
— One-line statement with no format, path, or scope detail7.261109.0
└──▷ ALSO FROM THESE RELEASES
Override the deduplication window for sighting relationships to 30 minutes while keeping the platform default for all other types.
A relationship type entry in relations_deduplication.types_overrides replaces the global configuration entirely: any past_days or next_days key omitted from the override resolves to 0, not the global default.
!
Invalid past_days or next_days values (e.g. 30m instead of PT30M) are silently ignored and produce a window of 0, with no validation error.
!
A relationship type override in relations_deduplication.types_overrides replaces the global configuration entirely: any key (past_days or next_days) absent from the override resolves to 0, not the global default.
!
Invalid duration values (e.g. 30m, which is neither a plain number nor a valid ISO 8601 duration) are silently ignored and produce a window of 0 — no validation error is raised.
!
The engine_check Elasticsearch configuration key is now ignored; replace it with elasticsearch:engine_client / ELASTICSEARCH__ENGINE_CLIENT.
Wazuh's latest updates focus on operational safeguards: memory thresholds that cap runaway rulesets in analysisd, queue-size warnings for the indexer connector and vulnerability scanner, new RBACRBACRole-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. read-only roles, event logging in plaintext and JSON, and Puppet module support for managing indexer credentials.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Wazuh - The Open Source Security Platform. Unified XDR and SIEM protection for endpoints and cloud workloads.
Wazuh's latest updates focus on operational safeguards: memory thresholds that cap runaway rulesets in analysisd, queue-size warnings for the indexer connector and vulnerability scanner, new RBACRBACRole-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. read-only roles, event logging in plaintext and JSON, and Puppet module support for managing indexer credentials.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
6 FEATURES0 OF 6 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Rule tree memory thresholds in analysisdnew95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
Set analysisd.rule_tree_memory_warning in the internal options file (or <rule_tree_memory_warning> in ossec.conf) to log a warning when the rule tree exceeds a memory threshold during ruleset loading; accepts a byte size (e.g.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ 2G) or a percentage of physical or cgroup memory (e.g. 25%), defaulting to 25%, with 0 disabling it. Set analysisd.rule_tree_memory_limit (same size/percentage format, disabled by 0 default) to reject a ruleset that would exceed a hard memory cap, blocking uploads via the Wazuh server API and dashboard with a descriptive error.
Cap rule tree memory on a container-based deployment so that an oversized ruleset upload is rejected before it exhausts cgroup memory.
Prevent an oversized ruleset from consuming all available memory by capping rule-tree growth and getting an early warning before the hard limit is hit.
— Names exact config keys, formats, defaults and a runnable config example.v4.14.10
0200
Puppet module support for indexer credentialsbreaking80how completely this was documenteddepth30/40specificity25/30actionability25/3080 / 100
The wazuh::manager Puppet class gains vulnerability_indexer_password and vulnerability_indexer_username parameters to configure the Wazuh manager keystore on every Puppet run and to override the default indexer account. Because the password is written to the keystore on every run, it must match the current indexer password or the manager becomes misconfigured after each run.
Set the indexer password and a custom indexer username in your Puppet manifest so the Wazuh manager keystore stays in sync on every run.
puppet
class { 'wazuh::manager':
vulnerability_indexer_password => 'MySecretPass123',
vulnerability_indexer_username => 'custom-indexer-user',
}
— Names exact parameters, class, and a runnable manifest, plus the breaking caveat.product docs
0300
Read-only RBAC roles for Reporting, Notifications, Snapshot Managementnew65how completely this was documenteddepth20/40specificity25/30actionability20/3065 / 100
Map users to the new reports_read_access, notifications_read_access, or snapshot_management_read_access roles in the Wazuh RBAC interface to grant read-only access to Reporting, Notifications, and Snapshot Management without exposing dashboard write permissions (avoid mapping to kibana_user or kibanauser backend roles).
Grant a read-only user access to Reporting, Notifications, and Snapshot Management without exposing dashboard write permissions.
📍Map the user to the following roles in the Wazuh RBAC interface:
- reports_read_access
- notifications_read_access
- snapshot_management_read_access
Do NOT map to: kibana_user role or kibanauser backend role.
— Names exact roles and UI steps but no deeper mechanism.product docs
thinner coverage below
0400
TCP session counter in wazuh-remotednew50how completely this was documenteddepth20/40specificity20/30actionability10/3050 / 100
wazuh-remoted statistics now include a tcp_sessions_unassociated counter reporting TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it. connections not yet associated with an agent.
— Names the exact stat field but no usage guidance.v4.14.10
0500
Queue-size warnings for indexer connector and vulnerability scannernew50how completely this was documenteddepth20/40specificity20/30actionability10/3050 / 100
Wazuh now logs a warning when a queue of the indexer connector or of the vulnerability scanner holds 100,000 elements or more, and again each time its size doubles.
— Gives concrete threshold and repeat behaviour but no config surface.v4.14.10
0600
Plaintext and JSON event logging formatsnew40how completely this was documenteddepth15/40specificity15/30actionability10/3040 / 100
Store logs received from third-party applications and systems via event logging in either plaintext or JSON format; both are disabled by default.
— States what the formats do but lacks config detail or commands.product docs
└──▷ BREAKING ON UPGRADE
!
The wazuh::manager class now writes vulnerability_indexer_password to the Wazuh manager keystore on every Puppet run; the value must match the current indexer password or the manager will be misconfigured after each run.
Databricks is a cloud data lakehouse platform that unifies data engineering, analytics, machine learning, and AI workloads on Apache Spark.
Databricks shipped a wide range of updates this window: AI Function tasks now run natively inside Lakeflow Jobs, Designer gained a mechanism-rich Multi-Row Formula operator and file export to Google Drive/SharePoint, Genie expanded with Ontology (GA), Clean Rooms, Volumes support and Slack/Teams approvals, and Unity Catalog added volume recovery (UNDROP VOLUME) and native Secrets management, alongside Asset Bundles' new sync_mode deployment controls and Unity Gateway's desktop AI agent provisioning.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
29 FEATURES2 OF 29 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Multi-Row Formula operator in Designernew100how completely this was documenteddepth40/40specificity30/30actionability30/30100 / 100
Designer adds a Multi-Row Formula operator that computes new columns from neighboring rows using shared Sort and Grouping fields to control row windows and group boundaries.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ It supports Running aggregate (Sum, Average, Minimum, Maximum, Count), Row number / rank (sequential, gap-rank, dense-rank, with Start at), Value from another row (lag/lead via Rows away and If no such row), Difference from previous row (via Rows back), Moving average over a configurable Window, Exponential smoothing with a Smoothing factor (α, default 0.3), and Custom expression for arbitrary Spark SQL OVER window expressions.
Compute a per-customer running total and row-over-row change in a visual workflow without writing SQL.
📍In the Designer canvas, add a Multi-Row Formula operator after your source, then:
1. Click Add formula › Running aggregate. Set New column name to customer_running_total, Function to Sum, Of column to amount. Add order_date (Ascending) as Sort and customer_id as Grouping.
2. Click Add formula › Difference from previous row. Set New column name to amount_change, Of column to amount, Rows back to 1. Same Sort and Grouping.
3. Click Apply.
— Every formula mode named with its controls and a runnable walkthroughproduct docs
0200
`workspace.sync_mode` deployment control in Asset Bundlesnew95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
Control how bundle files reach deployed resources with the new workspace.sync_mode config key (from version 1.23.0), accepting overwrite, in-place, or immutable — defaulting to in-place for development deployments from the workspace and overwrite otherwise. Setting workspace.sync_mode: immutable uploads bundle files and artifacts as a single immutable snapshot to block non-admin edits, recoverable with databricks bundle deploy --force after an out-of-band change; workspace.sync_mode: in-place references workspace source files directly without uploading. This replaces the deprecated presets.source_linked_deployment: true setting (use in-place instead of true, overwrite instead of false).
Prevent non-admin users from modifying deployed production job files by using an immutable snapshot deployment.
yaml
sync_mode: immutable
Force a fresh immutable snapshot after a snapshot was modified outside the bundle, unblocking the next deployment.
$ databricks bundle deploy --force
— Names every mode, default, recovery command and migration pathproduct docs
0300
UNDROP VOLUME and dropped-volume recoverynew95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
Restore a dropped volume with UNDROP VOLUME and list dropped volumes with SHOW VOLUMES DROPPED. On Databricks Runtime 19 or above, recover a dropped Unity Catalog volume within a 7-day window by locating its ID via SHOW VOLUMES DROPPED and restoring it — along with its original privileges and tags — using UNDROP VOLUME WITH ID.
Recover a volume accidentally dropped, then confirm it is no longer listed as dropped.
$ SHOW VOLUMES DROPPED;
UNDROP VOLUME <volume_name>;
Restore an accidentally dropped Unity Catalog volume, preserving its original privileges and tags, within the 7-day recovery window.
$ SHOW VOLUMES DROPPED;
-- Note the volume_id from the output, e.g. 'abc123'
UNDROP VOLUME WITH ID 'abc123';
`UCDataset` for loading Unity Catalog tables into PyTorchnew85how completely this was documenteddepth35/40specificity30/30actionability20/3085 / 100
Load Unity Catalog tables directly into PyTorch training loops on AI Runtime using UCDataset, with no Spark or SQL warehouse required. Apply filter expressions and deletion vectors at read time so only matching rows are streamed, and UCDataset automatically splits matching rows across GPUsGPUsA graphics processing unit, a processor that performs many calculations in parallel. AI workloads use GPUs to speed up model training and inference. and data loader workers (Beta).
Desktop AI agent provisioning via Unity Gatewaynew80how completely this was documenteddepth30/40specificity25/30actionability25/3080 / 100
Configure Claude desktop and ChatGPT desktop work agents with governed models, 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 skills through Unity Gateway, with subscription relay available in Claude desktop. Provision the Claude desktop app machine-wide via MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies. software using ug configure --global (macOS only), letting users import their claude.ai projects and chats.
Deploy a governed Claude desktop configuration to all managed Macs in your org without requiring users to configure it themselves.
Secrets in Unity Catalognew80how completely this was documenteddepth30/40specificity25/30actionability25/3080 / 100
Databricks now supports Secrets in Unity Catalog, letting you create secrets, grant access, and read them in notebooks, jobs, and UDFs. Grant broad secrets discoverability in a catalog — without exposing secret values or allowing code references — using GRANT BROWSE ON CATALOG <catalog_name> TO <principal>;.
Grant broad secrets discoverability in a catalog without exposing secret values or allowing code references.
$ GRANT BROWSE ON CATALOG <catalog_name> TO <principal>;
— Names usage surfaces plus a runnable GRANT statementproduct docs
0700
User-aware workspace instructions in Genienew75how completely this was documenteddepth25/40specificity25/30actionability25/3075 / 100
Customize Genie chat behavior per user group by adding group-specific logic to .genie_workspace_instructions.md, including a Load the user's info instruction before group-specific rules.
Route Finance and Sales users to different metrics in Genie chat responses without changing their data permissions.
markdown
# User-aware behavior
- Load the user's info to get their group memberships.
- If the user belongs to the Finance group, include gross margin and operating expenses in financial summaries.
- If the user belongs to the Sales group, include bookings and pipeline in performance summaries.
— Names the exact config file and shows a working instruction blockproduct docs
0801
AI Functions runtime requirement raised to DBR 17.3 LTSbreaking75how completely this was documenteddepth30/40specificity25/30actionability20/3075 / 100
Starting January 2027, AI Functions that previously required DBR 15.4 LTS or above (except ai_similarity) will require Databricks Runtime 17.3 LTS or above when called from SQL queries; calls on runtimes below 17.3 return an error, while ai_similarity and the AI Functions 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. are unaffected.
— Clear migration date and version threshold, no command neededproduct docs
0900
Okta connector in Lakeflow Connectnew75how completely this was documenteddepth30/40specificity30/30actionability15/3075 / 100
A new Okta connector in Lakeflow Connect (Beta) ingests identity and access data — users, groups, group memberships, app integrations, devices, policies, and administrator roles — using SSWS API token authentication with incremental ingestion.
— Lists all ingested object types and the auth methodproduct docs
1000
History control on shared views via ALTER SHAREnew75how completely this was documenteddepth25/40specificity25/30actionability25/3075 / 100
Control history sharing per view by adding WITH HISTORY or WITHOUT HISTORY to ALTER SHARE ADD VIEW and ALTER SHARE ALTER VIEW statements; existing history defaults are unchanged.
Explicitly exclude history when sharing a view in a Delta Sharing share, so recipients see only the current snapshot.
$ ALTER SHARE my_share ADD VIEW my_catalog.my_schema.my_view WITHOUT HISTORY;
— Exact SQL syntax and a runnable example givenproduct docs
1100
Date arithmetic in Lakeflow Jobs parametersnew75how completely this was documenteddepth25/40specificity25/30actionability25/3075 / 100
Shift a date in Lakeflow Jobs parameters by a duration before formatting it using dynamic value references such as {{(job.start_time - 1.day).iso_date}} (Beta).
Pass yesterday's date to a task parameter — useful for incremental ETL jobs that should process the previous day's data.
AI Function task type in Lakeflow Jobsnew70how completely this was documenteddepth30/40specificity25/30actionability15/3070 / 100
Lakeflow Jobs gains an AI Function task type (Beta) that runs ai_parse_document, ai_extract, or ai_classify on volume files or table rows without writing SQL; each run processes only inputs that are new or changed since the previous run.
— Names functions and incremental behavior, no exact config shownproduct docs
1300
Workflow export to Google Drive and SharePointNew integrationnew70how completely this was documenteddepth20/40specificity25/30actionability25/3070 / 100
Write Designer workflow output files to Google Drive or SharePoint by setting Destination to Google Drive or Microsoft 365, entering a Folder URL, and signing in via the connector.
Export workflow results to a SharePoint folder for sharing with stakeholders who don't have Databricks access.
📍In the Designer canvas, connect your final operator's outgoing port to a Destination operator. Set Destination to Microsoft 365, enable the SharePoint connector and click Connect. Paste your SharePoint folder link into Folder URL, enter a File name, and choose Excel as the format. Click Apply.
— Exact fields and a step-by-step example make it directly usableproduct docs
1402
MCP connectors enabled by default for compliance workspacesNew integrationbreaking65how completely this was documenteddepth25/40specificity25/30actionability15/3065 / 100
Databricks-provided MCP connectors in Agent (Google Drive, Gmail, Google Calendar, Microsoft 365, Atlassian, Slack, and GitHub) will be enabled by default starting December 2026 for workspaces with the compliance security profile enabled.
— Names all affected connectors and the exact trigger dateproduct docs
thinner coverage below
1500
GitHub Action skill publishing to Unity Gatewaynew55how completely this was documenteddepth20/40specificity20/30actionability15/3055 / 100
Publish Git-managed skills to Unity Gateway using a GitHub Action (beta): map folders to skills and trigger updates on merges to your publishing branch.
— Describes the mechanism but no exact action name or configproduct docs
1600
Moonshot AI Kimi K3.1 Pro on Unity Gatewaynew55how completely this was documenteddepth20/40specificity25/30actionability10/3055 / 100
Unity Gateway now supports Moonshot AI Kimi K3.1 Pro (with vision, reasoning, and agentic tool usetool useA model's ability to answer with a structured request to run a named function or API, which the application executes and returns. It is how agents take actions.) on pay-per-token endpoints in us-east-1.
— Names model, capabilities and region but no usage stepsproduct docs
1700
AI Functions spend tracked in Unity Gateway budgetsimproved55how completely this was documenteddepth20/40specificity25/30actionability10/3055 / 100
Unity Gateway budgets now include spend from AI Functions such as ai_parse_document, ai_extract, and ai_classify across batch and real-time execution paths.
— Names the tracked functions but no budget config shownproduct docs
1800
RocksDB default state store in Runtime 17.3 LTSimproved55how completely this was documenteddepth25/40specificity20/30actionability10/3055 / 100
Databricks Runtime 17.3 LTS and above now defaults Structured Streaming to the RocksDB state store provider, with changelog checkpointing also enabled by default.
— Names the default change but no opt-out or tuning detailproduct docs
1900
Genie Ontology (GA)new55how completely this was documenteddepth25/40specificity20/30actionability10/3055 / 100
Genie Ontology is now generally available: a business-aware context layer for Genie One, Genie Code, and other AI agents, combining Unity Catalog semantics you define with knowledge Genie extracts from your assets and usage patterns.
Ingress allowlist for BI tools and Genie chat appsnew50how completely this was documenteddepth15/40specificity20/30actionability15/3050 / 100
Allow traffic from Power BI, Tableau Cloud, dbt Cloud, and the Genie apps for Slack and Microsoft Teams in context-based ingress policies.
— Names all allowed tools but not the policy syntaxproduct docs
2100
Power BI semantic models as Genie knowledge sourcenew50how completely this was documenteddepth20/40specificity15/30actionability15/3050 / 100
Pull Power BI semantic model metadata into Genie as external knowledge sources (Beta), enabled by an account admin from the account console.
— Names the admin entry point but no detailed stepsproduct docs
2200
Read replicas for OLTP Data API readsnew45how completely this was documenteddepth20/40specificity15/30actionability10/3045 / 100
Databricks OLTP now supports read replicas, letting you offload Data API read requests from primary compute by sending them to a read replica's URL.
— Names the capability but gives no setup steps or limitsproduct docs
2300
Genie Clean Room (Beta)new45how completely this was documenteddepth20/40specificity15/30actionability10/3045 / 100
A new Genie Clean Room (Beta) collaboration type lets a provider share a curated Genie Agent with one collaborator, who can ask it questions in natural language without accessing the underlying data, context, or proprietary IP.
— Describes the model but no setup or API detailproduct docs
2400
Cross-workspace upstream job triggers in Lakeflownew45how completely this was documenteddepth20/40specificity15/30actionability10/3045 / 100
Trigger a Lakeflow Job when one or more upstream jobs complete, including across workspaces attached to the same metastore (Beta).
— Describes the trigger condition but no setup pathproduct docs
2500
Expandable visualization side panel in Genienew40how completely this was documenteddepth15/40specificity10/30actionability15/3040 / 100
Open any visualization in a side panel at larger size by clicking the expand control in the lower-right corner of the visualization card.
— Gives exact UI location but no further mechanismproduct docs
2600
Genie tool action approvals from Slack and Teamsnew40how completely this was documenteddepth15/40specificity15/30actionability10/3040 / 100
Approve or deny Genie tool actions directly from Slack and Microsoft Teams approval prompts without leaving the chat (Public Preview).
— States the capability with no configuration detailproduct docs
2700
Named feature flags in Asset Bundlesnew40how completely this was documenteddepth15/40specificity15/30actionability10/3040 / 100
Define named feature flags for a bundle using the new features mapping, where each key is the name of a feature (available from version 1.22.0).
— Names the config key but no usage exampleproduct docs
2800
Genie support for Volumesnew40how completely this was documenteddepth15/40specificity15/30actionability10/3040 / 100
Databricks Genie now supports Volumes so users can ask questions about unstructured files alongside structured data (Public Preview).
— States the capability with no mechanism detailproduct docs
2900
Tableau and Sigma as planned Genie knowledge sourcesnew20how completely this was documenteddepth10/40specificity10/30actionability0/3020 / 100
Databricks plans to support Tableau and Sigma as external knowledge sources for Genie.
— Roadmap mention only, no mechanism or timelineproduct docs
└──▷ ALSO FROM THESE RELEASES
└──▷ BREAKING ON UPGRADE
!
Beginning January 2027, AI Functions that previously required DBR 15.4 LTS or above (except ai_similarity) now require DBR 17.3 LTS or above; calls on runtimes below 17.3 return an error.
!
AI Functions in SQL queries will no longer be supported on Databricks Runtime versions below 17.3 LTS starting January 2027; ai_similarity and the AI Functions REST API are unaffected.
!
presets.source_linked_deployment is deprecated; set workspace.sync_mode: in-place instead of source_linked_deployment: true, or workspace.sync_mode: overwrite instead of source_linked_deployment: false.
!
AI Functions will require Databricks Runtime 17.3 LTS or above (an upcoming change from the current 15.4 LTS minimum).
Databricks adds Genie Search (GA), indexing org knowledge across connected sources with permission-aware results inside Genie One.
Databricks adds Genie Search, a generally available knowledge-indexing capability that connects multiple sources so Genie One can search all indexed content in one place.
Genie Search indexes each document's permissions alongside its content, so users see only documents they can already access in the source system.
Genie One now includes an org-directory feature powered by Genie Search that lets users find teammates, browse the org chart, and view profiles sourced from Workday.
Genie One's Discover view now supports searching Databricks assets and content from connected sources in a single search experience via Genie Search.
Trend Vision One shipped new Connection Policy controls for routing Smart Protection Services traffic, custom rule authoring across Integrity Monitoring, Intrusion Prevention, and Log Inspection policy resources, and richer attack-mode and impact visibility for AI Scanner findings.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Trend Vision One is Trend Micro's unified cybersecurity platform that consolidates threat detection, response, and risk management across endpoints, cloud, email, and networks.
Trend Vision One shipped new Connection Policy controls for routing Smart Protection Services traffic, custom rule authoring across Integrity Monitoring, Intrusion Prevention, and Log Inspection policy resources, and richer attack-mode and impact visibility for AI Scanner findings.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
3 FEATURES1 OF 3 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0101
Connection Policies route Smart Protection Services trafficnew95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
Runtime Connection Policy controls under Endpoint Security → Endpoint Security Configuration → Connection Policies let administrators route Smart Protection Services traffic through selected Service Gateways, any available Service Gateway, or directly to Trend Vision One.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ This requires agent version 202610 or later and a Service Gateway running Smart Protection Services 2.0.11 or later; enrollment applies to ALL agents assigned a connection policy and cannot be undone, so policies must be configured before enrolling.
Configure agent traffic routing before enrolling in Smart Protection Services to avoid an irreversible policy state.
📍In the Trend Vision One console, go to Endpoint Security → Endpoint Security Configuration → Connection Policies, open or create a runtime connection policy, set the Smart Protection Services connection method to selected Service Gateways, 'any available Service Gateway', or 'directly to Trend Vision One', then enroll in Smart Protection Services.
Attack mode and impact columns in AI Scanner findingsimproved78how completely this was documenteddepth30/40specificity28/30actionability20/3078 / 100
The AI Scanner's Evaluation results table gains a Mode column showing which attack mode (Managed, Custom, or Agentic) produced each finding, also displayed beside the attack objective in Scan details, plus an Impact column with plain-language labels such as Data disclosed: High, Data tampered: High, or Service disrupted: Low. The Objective filter was updated to group objectives under each attack mode, so selecting a mode and an objective returns only that mode's findings.
Narrow a review session to only findings produced by the Agentic attack mode so you can triage that mode's results in isolation.
📍In Trend Vision One, go to AI Security → AI Application Security → Evaluation results. Open the Objective filter, select Agentic, then choose the specific objective to display only Agentic-mode findings. Check the Mode and Impact columns to confirm the attack mode and plain-language impact label (e.g. Data disclosed: High) for each row.
Custom rule authoring across Endpoint Security policy resourcesnew75how completely this was documenteddepth30/40specificity27/30actionability18/3075 / 100
Endpoint Security Policies gain custom rule creation across three Policy Resources areas, all reachable via Endpoint Security → Endpoint Security Configuration → Policy Resources: custom Integrity Monitoring Rules that monitor registry values or files, or use custom XML (with support to duplicate, delete, and customize/reset TrendAI™ Integrity Monitoring rules); custom Intrusion Prevention Rules built from new custom application types (with support to configure/reset TrendAI™ Intrusion Prevention rules); and custom Log Inspection Rules authored in Basic format or custom XML targeting a specific log file path.
Create a custom rule to watch a sensitive registry value for unauthorized changes in an Endpoint Security Policy.
📍In the console, go to Endpoint Security → Endpoint Security Configuration → Policy Resources → Integrity Monitoring Rules, select Create Custom Rule, choose Registry Value as the monitor type, enter the target key and value name, then save and assign the rule to an Endpoint Security Policy.
Author a custom prevention rule to block traffic matching a known malicious pattern, assigned to the Advanced ruleset.
📍In the console, go to Endpoint Security → Endpoint Security Configuration → Policy Resources → Intrusion Prevention Rules, click Create, define your rule using a signature or start/end patterns, set the action to Prevent, and save. Assign the rule (Advanced ruleset) to the target Endpoint Security Policy.
When you need to inspect a custom application log file not covered by default rules, create a tailored Log Inspection rule scoped to that file.
📍In the console, go to Endpoint Security → Endpoint Security Configuration → Policy Resources → Log Inspection Rules, then select Create and choose Basic format or custom XML, specify the target log file path, and save the rule for use in an Endpoint Security Policy.
Smart Protection Services enrollment applies to ALL agents assigned a connection policy and cannot be undone — configure connection policies before enrolling.Our best guess
Rustinel is an endpoint detection for Windows, Linux, and macOS. Sigma, YARA, and IOC rules on native telemetry.
Rustinel v1.11.0 adds agent tamper and lifecycle alerting, Windows code-signing verdicts on image and driver loads, broader Windows telemetry collection, wider SigmaSigmaAn open, vendor-neutral rule format for describing patterns in log data, maintained by the Sigma project community, giving cyber tools a shared detection language that converts into platform-specific queries. rule coverage, and DNS event enrichment.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
6 FEATURES0 OF 6 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Windows code-signing verdicts on image loadsnew85how completely this was documenteddepth35/40specificity30/30actionability20/3085 / 100
Windows image and driver loads now expose Signed, Signature, and SignatureStatus fields with Sysmon semantics, covering embedded signatures, catalog-signed files, and AppX and MSIXMSIXA Windows application packaging format from Microsoft that bundles an app with its dependencies and metadata for clean install, update, and removal without registry pollution or file conflicts. package files.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Image loads now also verify AppX and MSIX package files against their signed block map.
— Exact field names and verification mechanism givenv1.11.0
0200
Expanded Windows telemetry fields and default eventsnew80how completely this was documenteddepth25/40specificity30/30actionability25/3080 / 100
Process starts now populate LogonId on both Windows and Linux, and CurrentDirectory on Windows. Windows kernel image loads are now classified as driver_load events. Security event 4740 is now collected by default on Windows, and Security 5379 joins the default subscription. New options windows.security_domain_controller_events (domain-controller events) and windows.security_long_tail_events (additional long-tail events) enable extra Windows Security event collection, both off by default.
Agent tamper detection and lifecycle alertingnew78how completely this was documenteddepth30/40specificity28/30actionability20/3078 / 100
A new agent::tamper::etw_provider_modified alert fires when an ETWETWEvent Tracing for Windows, a Microsoft kernel-level logging framework that lets applications and drivers emit structured events; cyber tools use it to observe process, network, and file activity without installing additional sensors. provider is disabled while the sensor is running. Loss alerts now also cover ETW sessions, eBPFeBPFA Linux technology for running verified programs inside the kernel without changing its source. Tools use it to observe processes, enforce rules and process network traffic. rings and maps, macOS ESF and BPFBPFA Linux kernel virtual machine that lets programs attach small, sandboxed bytecode filters to kernel hooks — syscalls, network events, tracepoints — so cyber tools can observe or redirect kernel behaviour without writing a kernel module. drops, and artifact queues. Alerts tagged event.dataset: "edr.agent" route lifecycle, tamper, and loss alerts to a separate pipeline that bypasses deduplication and active response.
— Names exact alert id and routing field, mechanism explainedv1.11.0
0400
DNS event enrichment and eBPF captureimproved65how completely this was documenteddepth25/40specificity25/30actionability15/3065 / 100
DNS events gain derived fields including registered domain and name shape for use in Sigma rules, and Linux eBPF now captures DNS over TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it. and recvmmsg responses while flagging truncated (cut) names.
— Specific capture paths named, limited usage guidancev1.11.0
0500
Sigma rule coverage for antivirus and auditdimproved60how completely this was documenteddepth20/40specificity25/30actionability15/3060 / 100
Sigma rule support now extends to category: antivirus rules, which are routed to Defender detections, and linux/auditd rules, which are satisfied through an auditd field view.
— Names categories but little mechanism detailv1.11.0
thinner coverage below
0600
SigmaHQ compatibility pin updatebreaking50how completely this was documenteddepth15/40specificity20/30actionability15/3050 / 100
The SigmaHQSigmaHQAn open community repository on GitHub that maintains the Sigma generic signature format and its growing library of detection rules, giving cyber tools a vendor-neutral rule set they can convert to any SIEM query language. compatibility pin moves to c04ad420, so sigma doctor verdicts on the SigmaHQ rule tree may change after upgrade.
— Names exact pin and affected command, thin explanationv1.11.0
└──▷ BREAKING ON UPGRADE
!
The SigmaHQ compatibility pin moves to `c04ad420`, so `sigma doctor` verdicts on the SigmaHQ rule tree may change after upgrade.Our best guess
Rapid7 InsightVM is a vulnerability management platform that scans networks and assets to identify, prioritize, and help remediate security weaknesses.
Defectdojo is an open-source Unified Vulnerability Management, DevSecOps & ASPM
DefectDojo 3.5.100 adds a system-wide read-only mode for the platform.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
1 FEATURE0 OF 1 FEATUREmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
System-wide read-only modenew55how completely this was documenteddepth15/40specificity20/30actionability20/3055 / 100
DefectDojo can be put into a system-wide read-only mode, configurable via settings.dist.py or local_settings.py.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
— Names config files but not the exact setting key or scope3.5.100
OSV-Scanner is a vulnerability scanner written in Go which uses the data provided by https://osv.dev
OSV-Scanner adds time-bounded vulnerability ignores via EffectiveUntil, warnings for stale [[PackageOverrides]], a Fix available HTML report filter, and ecosystem-priority severity in table output.
└──▷ GET THIS VERSIONDockerGogit
Docker
$ docker pull ghcr.io/google/osv-scanner:v2.7.1
Go
$ go install github.com/google/osv-scanner/v2/cmd/osv-scanner@v2.7.1
git
$ git clone --branch v2.7.1 https://github.com/google/osv-scanner.git
# already have the repo? check out this version:$ git checkout v2.7.1
published under the project's GitHub account · Go module path matches the repo · checked 10 Nov
Warn about unused or expired [[PackageOverrides]] configuration entries so stale overrides do not silently persist.
Use EffectiveUntil in [[IgnoredVulns]] configuration entries to set expiry dates on vulnerability ignores; IgnoreUntil is now deprecated.
Add a Fix available filter checkbox to HTML reports.
Now shows ecosystem-priority severity (such as Ubuntu severity) in table output when available.
Enable the os/ubuntu/binarytosource enricher in the vulns preset to map Ubuntu binary package names from Chisel manifests to source package names.
+2 moreshow less
Add line number support to the dotnet/paketlock extractor so findings reference exact file positions.
Support .tgz archive files in the embeddedfs/archive extractor.
The `haskell/cabal` extractor is renamed to `haskell/cabalprojectfreeze`; any configuration or scripts referencing it by name must be updated.Our best guess
The `chrome/extensions` extractor is renamed to `browser/extensions`; any configuration or scripts referencing it by name must be updated.Our best guess
Censys is a internet-wide scanning and search platform that catalogs hosts, services, and certificates to help security teams assess attack surface and exposure.
Censys introduced a PQCPQCCryptography designed to stay secure against a large quantum computer, which could break today's RSA and elliptic-curve keys. NIST standardized the first algorithms in 2024 as ML-KEM, ML-DSA and SLH-DSA. label to flag post-quantumpost-quantumCryptography designed to stay secure against a large quantum computer, which could break today's RSA and elliptic-curve keys. NIST standardized the first algorithms in 2024 as ML-KEM, ML-DSA and SLH-DSA.-cryptography-capable services and web properties, and significantly expanded its fingerprint database with over 200 new software, 40 new hardware, and 7 new OS fingerprints.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
2 FEATURES0 OF 2 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
PQC label for post-quantum cryptography assetsnew65how completely this was documenteddepth25/40specificity20/30actionability20/3065 / 100
Censys now adds a PQC label to services and web properties on the Platform that support post-quantum cryptographypost-quantum cryptographyCryptography designed to stay secure against a large quantum computer, which could break today's RSA and elliptic-curve keys. NIST standardized the first algorithms in 2024 as ML-KEM, ML-DSA and SLH-DSA., letting teams query and track PQC-capable assets across their attack surfaceattack surfaceThe exposed systems, interfaces and access points an attacker could try to use. It includes internet-facing services as well as identities, software and other reachable resources..︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Users can filter asset search or attack surface views by the PQC label to list all identified services and web properties.
Find all internet-exposed services in your attack surface that support post-quantum cryptography using the new PQC label.
📍Steps: In the Censys Platform, open the asset search or attack surface view and filter by label PQC to list all services and web properties identified as supporting post-quantum cryptography.
Expanded software, hardware, and OS fingerprint coveragenew60how completely this was documenteddepth20/40specificity30/30actionability10/3060 / 100
Censys adds over 200 new software fingerprints covering products such as Zabbix, CyberArk Privilege Cloud, CyberArk Privileged Access Manager Self-Hosted (PVWA), Palo Alto Networks Prisma Cloud Compute Console, FortiSOAR, Strapi 5, NGINX Plus, NGINX Unit, Cribl Stream, Apache Kafka, Apache Cassandra, etcd, Openfire XMPP Server, OPNsense, mySCADA myPRO HMI Runtime, Delta Electronics DIAEnergie, Delta Electronics DIAView, SAProuter, Rejetto HTTP File Server (HFS 2.x), and Salt (SaltStack) salt-api rest_cherrypy. It adds over 40 new hardware fingerprints for industrial control, networking, and OT devices including ABB AC500 V2 PLC CPUCPUThe main general-purpose processor in a computer. It runs application instructions and coordinates work with memory and other hardware., Phoenix Contact PLCnext Control, Moxa EDREDREndpoint detection and response, monitoring computers for suspicious behavior and providing investigation and response controls, such as isolating a device. Series Secure Router, Hitachi Energy RTU500 Series RTU, Fortinet FortiAP Wireless Access Point, Cisco Catalyst series switches and access points, SMA inverter and controller products, Fronius Datamanager, Growatt ShineLink and ShineLan-X Datalogger, and Furuno FELCOM Ship Earth Station. It also adds 7 new OS fingerprints including Sierra Wireless AirLink OS, Phoenix Contact AXC ProConOS eCLR, Veeam JeOS Appliance, Juniper Junos OS (SNMPSNMPA UDP-based protocol for querying and managing network devices, standardized by the IETF. Cyber tools use it to pull interface stats, topology data, and device state from routers, switches, and servers at scale.), LANCOM LCOS WEBconfig, SMA ennexOS, and Victron Energy Venus OS.
N-able N-central is a secure endpoints and automate IT operations at scale. N-central delivers unified endpoint management with advanced RMM capabilities for complex environments.
N-able N-central adds a new Windows OS patching engine to Endpoint Remediation, running independently of PME, with policies and additional controls coming soon.
New supported platformAdds a new Windows OS patching engine in Endpoint Remediation for on-demand patching, running independently of the existing Patch Management Engine (PME) with no shared settings.
Both the new Windows OS engine and PME can run side by side without disabling either.
Upcoming policy support will cover automated approvals, schedules, reboot behaviour, and tag-based device targeting.
Iru is a device-centric platform that unifies security, IT, and compliance around devices and everything they touch.
Iru expanded its single release this window with Mac support for Privacy Library Items, an upgraded SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management. sign-in experience, and in-product migration of Android devices to a managed Google domain.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
3 FEATURES0 OF 3 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
In-product Android managed Google domain migrationnew40how completely this was documenteddepth15/40specificity15/30actionability10/3040 / 100
Iru now supports upgrading Android devices to a managed Google domain directly within the product.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Discover the latest features, improvements, and updates to Ory.
Ory tightened security controls this window, enforcing stricter 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. assertion validation and locking down token lifespan configuration for dynamically registered OAuth clients.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
2 FEATURES2 OF 2 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0101
SAML assertion issuer enforcement in Ory Polisbreaking55how completely this was documenteddepth25/40specificity20/30actionability10/3055 / 100
Ory Network (via Ory Polis) now accepts a SAML assertion only from the IdPIdPA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems. of its connection, requiring the Issuer of a signed SAML assertion to match the entity ID of the configured IdP.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Assertions from other issuers will be rejected.
Admin-only token lifespans for dynamic client registrationbreaking55how completely this was documenteddepth25/40specificity20/30actionability10/3055 / 100
Ory Hydra now restricts token lifespan configuration to administrators for dynamically registered clients, removing per-client token lifespan overrides from 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. Dynamic Client Registration. OpenID Connect Dynamic Client Registration no longer accepts per-client token lifespan settings.
OpenID Connect Dynamic Client Registration no longer accepts per-client token lifespan settings; token lifespans for dynamically registered clients are now administrator-only.Our best guess
Ory Polis now requires the Issuer of a signed SAML assertion to be the entity ID of the connection's IdP; assertions from other issuers will be rejected.Our best guess
Secure users, AI agents, and more with Auth0, a production-ready authentication and authorization platform built to help you ship faster and drive sign-ups.
Auth0 shipped GA Dashboard Search with a global Command Palette and Boolean filtering, Early Access Management API search endpoints for Clients, Client Grants, Resource Servers and Organizations, and Early Access support for securing Universal Commerce Protocol (UCP) servers so AI agents can shop on users' behalf.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
3 FEATURES0 OF 3 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Dashboard Search reaches general availabilityimproved95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
A global Command Palette (Cmd+K on macOS, Ctrl+K on Windows/Linux) now searches Applications, APIs, Organizations, and Users by name, display name, or ID from any Dashboard page.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Guided filters and Boolean search are available in the search bar of Applications, APIs, Roles, Custom Domains, Organizations, Organization members, Enterprise Connections, and Database Connections, combining up to 5 filters with AND/OR logic, searching metadata by key and value or key only, and using NOT to exclude results; filter pills persist in the URL so filtered views can be shared or bookmarked.
Find a specific Organization from any Dashboard page without navigating to the Organizations section first.
📍Press Cmd+K (macOS) or Ctrl+K (Windows/Linux) to open the Command Palette, then type the organization name, display name, or ID to jump directly to it.
Build a shareable filtered view of Enterprise Connections using multiple conditions and send it to a teammate.
📍Navigate to Dashboard › Authentication › Enterprise Connections, use the search bar to add up to 5 AND/OR filter pills (e.g. by name or metadata key/value), then copy the URL — the filters are encoded in it and will restore when opened.
Management API search for Clients, Client Grants, Resource Servers, Organizationsnew85how completely this was documenteddepth25/40specificity30/30actionability30/3085 / 100
New Early Access endpoints let you search directly instead of paging through full lists or filtering client-side: GET /api/v2/clients/search, GET /api/v2/client-grants/search, GET /api/v2/resource-servers/search, and GET /api/v2/organizations/search.
Find a specific client by name without iterating pages — useful in automation scripts that need to look up a client ID before provisioning.
$ curl -X GET 'https://<your-domain>/api/v2/clients/search?q=my-app-name' \
-H 'Authorization: Bearer <mgmt-api-token>'
Locate an organization by identifier during an incident response workflow without client-side filtering across paginated results.
$ curl -X GET 'https://<your-domain>/api/v2/organizations/search?q=acme-corp' \
-H 'Authorization: Bearer <mgmt-api-token>'
Auth for Universal Commerce Protocol (UCP)new45how completely this was documenteddepth25/40specificity15/30actionability5/3045 / 100
Auth0 now supports securing a UCP server in Early Access so trusted AI agents such as Gemini can connect to it and act on behalf of shoppers, including discovering catalogs and making purchases. User identities can be linked so a shopper's identity in an AI agent connects to their account with a business, enabling personalization, loyalty points, and order tracking.
1Password's biggest addition this window is Environments — shareable, per-project secret collections with matching support added across the CLI (op environment secret list), SDKs (list_secrets()), and the 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 (append_secrets/list_secrets), alongside a new Events API token-introspection endpoint and new environments/provisioning CLI command groups.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
All the latest releases from the 1Password team
1Password's biggest addition this window is Environments — shareable, per-project secret collections with matching support added across the CLI (op environment secret list), SDKs (list_secrets()), and the 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 (append_secrets/list_secrets), alongside a new Events API token-introspection endpoint and new environments/provisioning CLI command groups.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
6 FEATURES1 OF 6 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
`op environment secret list` commanddeprecated95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
The CLI adds op environment secret list { <environmentName> | <environmentID> | - } to list secrets in a 1Password Environment, returning each secret's name, value, and concealment status; output can be formatted with --format (or OP_FORMAT) as human-readable, json, csv, or dotenv, and concealed values can be revealed with --reveal.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Secrets are concealed in stdout/stderr by default; the CLI can also print a single named secret or run a script provisioned with Environment secrets directly. The prior command for listing Environment secrets is deprecated on November 17, 2026 and removed on January 20, 2027 — callers must migrate to op environment secret list.
Export all secrets from a staging environment into a dotenv file for local development, including concealed values.
$ op environment secret list Staging --reveal--format=dotenv > .env
Inspect all secrets in a specific environment by ID in CI, keeping concealed values hidden for safe audit logging.
$ op environment secret list 6ijsqfxvdxlwjcpsx7yrzs3tka --format=json
List all secrets in a specific 1Password Environment for a local development setup.
$ op environment secret list <environment-id>
— Full command, flags, env var and migration deadline givenproduct docs
0200
MCP server tools for Environment secretsnew77how completely this was documenteddepth32/40specificity25/30actionability20/3077 / 100
The 1Password MCP server adds append_secrets, to add or update secrets in an Environment (recommended for rotating secrets previously stored in plaintext in code or .env files), and list_secrets, which returns only secret names, never values. Claude Code can import an existing env file into a new Environment via the 1password-environments skill — it creates the Environment, mounts it at the same file path, and stores values (API keys, tokens, passwords, private keys, connection strings) as concealed secrets, prompting to merge or rename if a duplicate Environment name is found. On macOS 27 Golden Gate, an MCP error -32000: Connection closed is fixed by enabling the Data shared by 1Password permission for the client app under Privacy & Security > Files & Folders.
Rotate a leaked secret by overwriting it in a 1Password Environment rather than leaving it in a plaintext .env file.
📍"Add a secret for my OpenAI API key" (triggers append_secrets via your MCP client, e.g. Claude or Cursor)
Inspect which secrets are defined in a project environment before running a workflow that depends on them.
📍"Show me the secret names in my project environment" (triggers list_secrets via your MCP client)
Troubleshoot MCP connection failures on macOS 27 Golden Gate when the 1Password server reports MCP error -32000: Connection closed.
📍In System Settings, go to Privacy & Security › Files & Folders, find the Claude app (or your terminal/editor), and turn on the setting that starts with Data shared by 1Password. Quit and reopen the app, start a new Claude Code session, then check the server status.
Ask your MCP client to list what secrets exist in a project environment before adding new ones.
📍Show me the secret names in my project environment
Rotate a secret that was previously stored in plaintext — tell your MCP client to store the new value in a 1Password Environment.
📍Add a secret for my OpenAI API key
If your MCP client cannot connect after upgrading to macOS 27 Golden Gate, grant it file access to unblock the 1Password MCP server.
📍Open System Settings › Privacy & Security › Files & Folders, select your MCP client app (e.g. Claude, Cursor, or Codex), and turn on the setting that starts with Data shared by 1Password. Then quit and reopen your MCP client.
— Named tools and exact fix path, but only natural-language triggersproduct docs
0300
Events API token introspection endpointnew75how completely this was documenteddepth20/40specificity25/30actionability30/3075 / 100
The Events API adds a v2 token-introspection endpoint, GET https://events.1password.com/api/v2/auth/introspect, which inspects a bearer tokenbearer tokenA credential that grants access to whoever presents it, without a separate proof of possession. Protecting the token matters because a stolen copy can be used by someone else.'s validity and metadata when called with an Authorization: Bearer <token> header.
Verify that a 1Password Events API bearer token is still valid before issuing requests against the Events API.
— Exact endpoint, method, header and runnable curl exampleproduct docs
0400
SDK `list_secrets()` function for Environmentsdeprecated63how completely this was documenteddepth28/40specificity20/30actionability15/3063 / 100
1Password SDKs add a list_secrets() function to read secrets from an Environment directly — it accepts an Environment name or ID plus a reveal option and returns a list of EnvironmentSecret objects without the previous response wrapper. The existing secrets-reading function and the EnvironmentVariable types it returned are deprecated on November 17, 2026 and removed on January 20, 2027; callers must migrate to list_secrets().
— Function and types named but no runnable code example givenproduct docs
0501
Environments for organizing and sharing secretsnew60how completely this was documenteddepth30/40specificity15/30actionability15/3060 / 100
1Password introduces Environments — shareable, context-scoped secret collections for each project, application, or deployment stage (e.g. staging or production), replacing per-project .env file setup. Team members on shared accounts can be granted per-Environment access to keep secrets for different projects or stages isolated, and secrets on an Environment's details page can be searched or sorted alphabetically via the Search secrets field. Secrets can be added manually (desktop app New secret) or imported in bulk from an existing .env file, and secrets loaded from an Environment can be combined with secret references from a .env file in the same invocation via 1Password CLI or SDKs.
Set up a development environment's secrets and load them into a local script without writing plaintext secrets to disk.
📍Steps: 1. Open and unlock the 1Password desktop app. 2. Navigate to your project's Environment. 3. Select + Add secret, enter the secret name in the Name field and its value in the Value field. 4. Repeat for each secret. 5. Use 1Password CLI or an SDK to read and pass the secrets to your application as environment variables.
Add a secret manually to an Environment during local project setup when you don't have an existing .env file to import.
📍In the 1Password desktop app: open and unlock the app, navigate to your Environment, select New secret, enter the secret name in the Name field and its value in the Value field, then repeat for each secret.
Bootstrap a project environment quickly by importing all variables from an existing .env file instead of adding secrets one by one.
📍In the 1Password desktop app: open and unlock the app, navigate to your Environment, then select the option to import a file — 1Password will automatically import the environment variables from the file as secrets.
— UI and conceptual workflow described, no CLI/API flags named hereproduct docs
thinner coverage below
0600
New `environments` and `provisioning` CLI command groupsnew33how completely this was documenteddepth8/40specificity15/30actionability10/3033 / 100
The CLI reference adds two new top-level command groups: environments, for managing 1Password Environments and their secrets, and provisioning, for managing provisioning integration settings.
— Only names two command groups, no subcommands detailedproduct docs
The existing secrets-reading function and the response-wrapper types it returns will be deprecated on November 17, 2026 and removed on January 20, 2027; callers must migrate to list_secrets().
!
The existing command to list secrets in a 1Password Environment via 1Password CLI will be deprecated on November 17, 2026 and removed on January 20, 2027 — replace it with op environment secret list.
!
The existing function in 1Password SDKs and the EnvironmentVariable types it returns will be deprecated on November 17, 2026 and removed on January 20, 2027 — replace it with list_secrets().
1Password SaaSSaaSSoftware delivered over the internet and billed by subscription, hosted and operated by the vendor so the buyer skips installation and infrastructure management. Manager adds scoped budgets and spend limits, groups OpenAI usage records, adds human-readable Claude API key names, and expands log redaction.
Create scoped budgets and spend limits for applications, teams, cost centers, and API keys.
OpenAI request-level usage is now grouped into more useful records, reducing noise from individual request identifiers.
Claude usage imports now include human-readable API key names.
Jira-connected tasks can now be closed or deleted manually when the related Jira issue is closed or deleted.
Log redaction now covers additional secrets and preserves boolean values, and reduces sensitive provider, user, and vendor response data written to integration and workflow logs.
Aembit is a runtime IAM for AI agents and workloads. Enforce access policies, keep downstream credentials out of agents, audit access, and revoke it with a kill switch.
Aembit's centerpiece this window is 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. Identity Gateway, a managed (or self-hosted) proxy that enforces Access Policies and injects per-request credentials between AI agents and MCP servers, paired with blended user+agent identity policies, new MCP access reporting, Microsoft Copilot Studio support, SSOSSOSingle sign-on, using one identity provider to sign in to multiple applications. It reduces separate passwords and centralizes account and access management./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.-group controls on Client Workload identification, and a Windows 11 Aembit Desktop with a credential-free local proxy.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
7 FEATURES1 OF 7 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0101
MCP Identity Gateway for AI agent-to-MCP-server accessnew98how completely this was documenteddepth40/40specificity30/30actionability28/3098 / 100
Aembit adds MCP Identity Gateway, a stateless data-plane proxy reachable at https://<tenant-id>.mcpgateway.aembit.io that enforces Access Policies on every AI-agent-to-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. request and injects credentials per request so agents never hold downstream secrets; it can also be self-hosted on a Linux host you provision, configured via environment variables Aembit sets during provisioning, keeping all runtime state within one Aembit Tenant.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ The gateway validates a signed JWTJWTAn open standard (RFC 7519) for encoding claims as a signed, optionally encrypted JSON object passed between parties. Cyber tools use it to verify identity and permissions without querying a central session store. against Aembit's JWKSJWKSA JSON Web Key Set, a published collection of cryptographic keys. Applications use it to find the public key needed to verify a token's signature. signing keys on every request, caches downstream credentials per-request for 60 seconds without persisting them to disk, denies all requests by default (fail-closed) unless a policy explicitly allows them, and produces structured logs of agent identity, user identity, target server, and policy decisions. Access is governed by two Access Policies — a left-side Client-to-Gateway policy and a right-side Gateway-to-Server policy — with optional time-based and 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.-based Access Conditions and a Content Security provider enforcing four boundaries (token validation against the Trust Provider, TLS termination/re-initiation, in-memory-only credential handling, and per-request policy evaluation); the gateway auto-discovers available MCP servers, proxies resource discovery/retrieval through the same flow as tools, and provides per-user credential isolation with centralized routing to multiple MCP servers from one endpoint.
Verify your tenant's MCP Identity Gateway endpoint is live and TLS is terminating correctly before configuring any client workloads.
Walk through the two required Access Policies and optional Content Security provider in the Aembit console to fully wire up an AI agent to an MCP server.
📍1. In the Aembit console, go to Access Policies and create the Client-to-Gateway policy: set the Client Workload to the MCP client, set the Server Workload hostname to <tenant-id>.mcpgateway.aembit.io, and attach a Credential Provider that issues tokens whose audience matches MCP Identity Gateway.
2. Create the Gateway-to-Server policy: set the Client Workload to MCP Identity Gateway (https://<tenant-id>.aembit.io), set the Server Workload to the upstream MCP server, and configure OAuth with the Authorization URL and Token URL from the MCP server vendor.
3. Optionally, open the Access Conditions card in the right panel on either policy and add time-based or GeoIP-based conditions.
4. Optionally, attach a Content Security provider to the Gateway-to-Server policy to inspect each MCP tool message.
Route an AI agent through MCP Identity Gateway instead of directly to an MCP server, so the agent authenticates through Aembit and holds no downstream credentials.
📍In your Aembit Tenant, go to MCP Identity Gateway › Access Policies and create a Gateway-to-Server policy targeting the MCP server you want to front. Point your AI agent at the MCP Identity Gateway endpoint in place of the direct MCP server URL.
— Mechanism, defaults and architecture fully detailed with runnable examples.product docs
0200
Aembit Desktop for Windows 11 with credential-free proxynew93how completely this was documenteddepth35/40specificity30/30actionability28/3093 / 100
Aembit Desktop is a new Windows 11 x64 (public beta) application for accessing HTTP and HTTPSHTTPSHTTP carried over an encrypted TLS connection. It protects traffic in transit and lets a browser verify the server's certificate. Server Workloads without static credentials, via a local proxy that injects service credentials into requests per-policy with no credentials stored on the machine. Administrators deliver managed Desktop settings as JSON files to override per-user configuration, and users can sign in through an existing SSO Identity Provider once Enable User Access is toggled on that Identity Provider's Users tab (visible only after Aembit support enables Desktop for the tenant). The adt beta tool aws command runs AWS CLI commands without storing AWS access keys on the machine, pulling credentials from an Access Policy at request time, e.g. adt beta tool aws s3 ls s3://my-bucket.
Let Windows 11 users sign in to Aembit Desktop through an existing SSO Identity Provider without static credentials.
📍In the Aembit console, go to Administration › Identity Providers › <your IdP> › Users tab, then toggle Enable User Access and save. The Users tab appears only after Aembit support enables Aembit Desktop for your tenant.
Run an AWS CLI command from a developer workstation without any stored AWS access keys — Aembit Desktop injects the credential from the matching Access Policy.
$ adt beta tool aws s3 ls s3://my-bucket
Run an AWS CLI command from a developer workstation without storing AWS access keys — the credential comes from the Access Policy at request time.
Microsoft Copilot Studio agent support in MCP Identity Gatewaynew90how completely this was documenteddepth35/40specificity30/30actionability25/3090 / 100
Connect Microsoft Copilot Studio agents to MCP Identity Gateway over Streamable HTTPStreamable HTTPAn MCP transport protocol that lets a client send JSON-RPC requests over HTTP POST while the server streams responses back using Server-Sent Events, replacing the older stdio and SSE transports for remote tool communication. transport, with the gateway validating 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 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. and proxying requests to downstream MCP servers using Aembit-managed credentials. For servers without OAuth discovery, such as Microsoft GraphMicrosoft GraphMicrosoft's unified REST API for accessing data and services across Microsoft 365, Azure AD, and related cloud products, giving cyber tools a single authenticated endpoint to query users, emails, files, and sign-in logs. via the Microsoft MCP Server for Enterprise, configure the callback URL manually as https://<tenant-id>.id.<stack>.aembit.io/mcp-auth/userauth/<cp-id>/callback. To stay within Copilot Studio's 70-tool-per-agent limit, assign fewer Access Policies per agent, use per-tool toggles, split integrations across child agents with their own tool budgets, or keep MCP tool names to 64 characters or fewer including the Server Workload prefix MCP Identity Gateway adds.
Connect a Copilot Studio agent to MCP Identity Gateway when the target MCP server (such as Microsoft Graph) does not support OAuth discovery and requires manual OAuth configuration.
📍In Copilot Studio, create a tool with your MCP Identity Gateway endpoint (ending in /mcp) as the Server URL, then when prompted for OAuth values enter the callback URL manually: https://<tenant-id>.id.<stack>.aembit.io/mcp-auth/userauth/<cp-id>/callback
— Exact callback URL format and concrete workaround limits given.product docs
0400
SSO enforcement and OIDC group matching for Client Workloadsnew90how completely this was documenteddepth35/40specificity30/30actionability25/3090 / 100
Client Workloads gain an Enforce SSO setting (available on both Redirect URI and CIMDCIMDA text-based Short Message Service protocol from Nokia for exchanging SMS messages between applications and SMS centers, used to build tools that send, receive, or test carrier messaging flows. Client ID identification) requiring users to authenticate through a configured SSO 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. during the OAuth flow — a multi-select dropdown lists active SSO providers and at least one must stay selected or the save is rejected; for CIMD Client ID workloads the toggle also limits the MCP sign-in page to only the assigned providers. A single Client Workload can now carry both a Redirect URI and a CIMD Client ID identifier at once. A new User Group identifier lets Access Policies match MCP Identity Gateway users by the groups they belong to in an OIDC Identity Provider (via User Groups & MCP Policy Access group-lookup settings on the provider), and Aembit now supports automatic user creation tied to Identity Providers at sign-in. Separately, the MCP Authorization Server now accepts an application_type of native or web via Dynamic Client Registration, allowing native clients to register custom-scheme redirect URIs such as vscode://, and MCP Server Workload URL Path values can now end in a trailing slash, such as /mcp/.
Enforce that all Client Workload OAuth flows go through a specific SSO identity provider, preventing unauthenticated or direct access.
📍In the Aembit console, go to Access Policies › Client Workloads › [your workload] › Identification › Redirect URI, enable Enforce SSO, then select at least one active SSO identity provider from the multi-select dropdown and save.
Disable SSO enforcement for a Client Workload to allow the OAuth flow to complete without user authentication — useful for fully automated, non-human workloads.
📍In the Aembit console, open the Client Workload, go to MCP Authorization Configuration, and set Enforce SSO to Off.
Restrict which identity providersidentity providersA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems. users can choose on the MCP authentication page for a specific Client Workload.
📍In the Aembit console, open the Client Workload, go to MCP Authorization Configuration, set Enforce SSO to On, then deselect any identity providers you do not want shown on the MCP authentication page.
Let Windows 11 users sign in to Aembit Desktop through an existing SSO Identity Provider without static credentials.
📍In the Aembit console, go to Administration › Identity Providers › <your IdP> › Users tab, then toggle Enable User Access and save. The Users tab appears only after Aembit support enables Aembit Desktop for your tenant.
Restrict an MCP server to users who belong to the engineering team or any group starting with finance- in your OIDC Identity Provider.
📍In the Aembit console, open the Client Workload, add a User Group identifier, select the OIDC 1.0 Identity Provider, and enter engineering or finance-* as the group value. Then go to the Identity Provider's Settings and enable User Groups & MCP Policy Access, setting the group claim name and cache time.
Restrict an MCP server to members of a specific engineering group defined in an OIDC Identity Provider, using wildcard patterns for teams like all finance sub-groups.
📍In the Client Workload editor, add a User Group identifier, select your OIDC 1.0 Identity Provider, and enter engineering or finance-* as the group pattern. Enable User Groups & MCP Policy Access on the Identity Provider's settings to set the group claim and cache time.
Enforce SSO for an MCP client identified only by CIMD Client ID, so only users from approved Identity Providers can reach the MCP server.
📍In the Client Workload editor, set the identifier to a CIMD Client ID, toggle on Enforce SSO, and select one or more Identity Providers. Save requires at least one Identity Provider selected.
— Every toggle, identifier and config value named with exact steps.product docs
0500
MCP access reporting and live authorization viewnew75how completely this was documenteddepth30/40specificity25/30actionability20/3075 / 100
Aembit surfaces MCP Access Activities reports, listing MCP requests with no error or exception code grouped by user, MCP client, MCP server, method, and target via the Reporting dashboard; MCP Access Exceptions reports list requests carrying an error or exception code grouped the same way plus exception type. A live authorization request view at MCP Identity Gateway surfaces the redirect URI, resource, matched Client Workload, and Access Policy outcome for each inbound request ahead of the standard event-pipeline delay, and every MCP request is logged with both verified user identity and AI agent workload identityworkload identityAn identity assigned to software, such as a service or container, rather than a person. It lets the workload authenticate and receive permissions without sharing a user's credentials., capturing authentication events (user, agent, time, location) and workload events (resources accessed, operations performed, credentials issued) for compliance audit trails.
Review failed or blocked MCP requests at a glance — go to the Reporting dashboard and open MCP Access Exceptions to see every request that returned an error or exception code, grouped by user, client, server, method, target, and exception type.
📍In the Aembit Tenant, go to Reporting › MCP reports › MCP Access Exceptions to view all MCP requests with an error or exception code, grouped by user, MCP client, MCP server, method, target, and exception.
Review failed or blocked MCP requests in one place to triage access policy issues or client misconfigurations.
📍In the Aembit console, go to Reporting › MCP Access Exceptions. Open any row to see the exception category and code, the Access Policy decisions applied, and the per-server outcome for that request.
— Named dashboards and fields with a clear navigation path.product docs
0600
Blended identity combining user and agent identitynew70how completely this was documenteddepth35/40specificity25/30actionability10/3070 / 100
Access policies can now evaluate blended identity — combining user identity from your IdPIdPA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems. with AI agent workload identity — so a single policy can ask whether this user, using this AI agent, is authorized to reach a resource right now, enabling per-agent-per-user rules such as allowing Confluence access through Claude Desktop but denying it through an untrusted third-party agent. The mechanism spans two Access Policies: a Credential Provider embeds the user's identity into the 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. via dynamic claim injection, and MCP Identity Gateway uses its workload identity plus the embedded user context to obtain downstream credentials, storing OAuth tokens per user rather than per agent so revoking one user's access doesn't affect others sharing the same gateway.
— Mechanism and scope explained, but no example command given.product docs
thinner coverage below
0700
MCP Server for AI-assistant tenant managementnew30how completely this was documenteddepth15/40specificity10/30actionability5/3030 / 100
Aembit adds an MCP Server that exposes Aembit Tenant configuration to MCP clients, so practitioners can query and manage Aembit from an AI assistant.
— Only a one-line description, no interface or command given.product docs
Aembit adds Google Antigravity support via the 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. Identity Gateway, brokering short-lived credentials using a blended agent-and-user identity model with full audit logging.
Point mcp_config.json at the Aembit MCP Identity Gateway to enforce real-time policy evaluation for Google Antigravity (agy) agents before any credential is issued.
Aembit's blended identity model evaluates both the agent's identity and the human user's IdPIdPA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems. identity together, replacing long-lived Google ADC sessions and bearer tokensbearer tokensA credential that grants access to whoever presents it, without a separate proof of possession. Protecting the token matters because a stolen copy can be used by someone else. stored in mcp_config.json or cached at ~/.gemini/antigravity/mcp_oauth_tokens.json with short-lived, scoped credentials.
Every credential issuance and access event is logged against both the agent identity and the human identity behind it, making 'which agent touched production, and on whose behalf' a query rather than an investigation.
Panorays, a Forrester Wave Leader, helps 500+ companies assess, monitor and fix third-party cyber risk based on business context.
Panorays rolled out 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 integration under Panorays Axis for AI/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 third-party risk workflows, broadened its public API around the questionnaire lifecycle, and added new widgets and filters to Live Dashboards.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
3 FEATURES0 OF 3 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
MCP server integration via Panorays Axisnew55how completely this was documenteddepth25/40specificity20/30actionability10/3055 / 100
Panorays Axis now connects AI/LLM tools to Panorays tenant data through a remote MCP server, with guided workflows for questionnaire analysis, assessment reports, and supplier onboarding.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ API documentation adds a dedicated section covering Panorays Axis (MCP).
— Describes mechanism and use cases but no exact endpoint or command givensnapshot-20261110
0200
Public API expansion for questionnaire lifecycleimproved45how completely this was documenteddepth20/40specificity15/30actionability10/3045 / 100
The public API now supports questionnaire lifecycle management, evidence validation retrieval, and expanded questionnaire data fields, with documentation adding dedicated sections for TPCRM lifecycle use cases and additional endpoint references under the Panorays domain.
— No specific endpoint paths, methods, or field names providedsnapshot-20261110
0300
New widgets and filters in Live Dashboardsimproved35how completely this was documenteddepth15/40specificity10/30actionability10/3035 / 100
Live Dashboards now include new interactive widgets, pre-configured dashboards, and expanded filtering options across Panorays data points.
— No specific widget names, dashboard templates, or filter fields listedsnapshot-20261110
Runlayer's 4.0.0 release adds 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. gateway multi-model failover with stricter routing-rule validation, new self-hosted secret management options, and a broad set of 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.-driven controls spanning agent administration, webhook configuration, poll triggers, and eval workflows.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Runlayer is an MCP gateway that enforces identity-mapped permissions and detects threats across the agents and MCP servers employees use.
Runlayer's 4.0.0 release adds 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. gateway multi-model failover with stricter routing-rule validation, new self-hosted secret management options, and a broad set of 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.-driven controls spanning agent administration, webhook configuration, poll triggers, and eval workflows.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
7 FEATURES0 OF 7 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
LLM gateway multi-model failover with routing-rule validationbreaking95how completely this was documenteddepth35/40specificity30/30actionability30/3095 / 100
LLM gateway routing rules can now set up to three fallback models so the gateway retries the next fallback after a server error, rate limit, timeout, or refused key, across providers if needed, configured via Settings → AI models → Budgets & routing.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Creating or updating a routing rule via the API or Terraform now returns a 422 error if the rule's target or any fallback isn't listed under your AI providers; add the model under Settings → AI models → Providers or pull a valid id from GET /api/v1/llm-gateway/models. Existing rules keep routing, and updates that leave targets unchanged still succeed.
Discover which LLM models are available to use as routing-rule targets or fallbacks before creating a rule.
Provisioned secret source for ECS deploymentsnew81how completely this was documenteddepth25/40specificity28/30actionability28/3081 / 100
Self-hosted ECSECSElastic Common Schema, an open specification maintained by Elastic that defines a standard field naming convention for log and event data, letting cyber tools share and correlate records without per-source field mapping. deployments can set tenant_secret_source = "provisioned" so WorkOS, Sentry, and Resend keys are read directly from your AWS Secrets Manager secret instead of passing through Terraform state.
Keep WorkOS, Sentry, and Resend secrets in AWS Secrets Manager only, preventing them from appearing in Terraform state on ECS deployments.
RESEND_API_KEY for self-hosted email deliverynew78how completely this was documenteddepth25/40specificity25/30actionability28/3078 / 100
Self-hosted deployments can supply the Resend API key as the RESEND_API_KEY environment variable or the runlayer-app Kubernetes secret key, which takes priority over the WorkOS Vault secret fallback.
Inject the Resend API key into a Kubernetes self-hosted deployment so email delivery does not depend on WorkOS Vault.
— Exact env var, secret name, and kubectl command given.4.0.0 (2026-11-09)
0400
Webhook signing and payload editing via MCPnew75how completely this was documenteddepth25/40specificity28/30actionability22/3075 / 100
Through the Runlayer Assistant or an MCP client, users can configure request signing, rotate or disable signing secrets, and rename agent webhooks. The webhook JSON body can also be edited in Settings → Policies → Approval settings using variables like {{request_id}}, {{session_id}}, and {{status}}, with a Send test option to try the payload.
— Names template variables and UI path for the editor.4.0.0 (2026-11-09)
thinner coverage below
0500
Agent administration controls via MCPnew55how completely this was documenteddepth20/40specificity20/30actionability15/3055 / 100
Through Runlayer MCP or the Runlayer Assistant, users can run a saved agent schedule once on demand without changing its next or last run time, transfer agent ownership to another person (subject to the same permission checks as the agent's Share dialog), and generate an AI icon suggestion for an agent that can be applied with update_agent.
— Names MCP tool and dialog but no full workflow detail.4.0.0 (2026-11-09)
0600
Poll trigger introspection via MCPnew53how completely this was documenteddepth18/40specificity20/30actionability15/3053 / 100
Through Runlayer MCP, users can list a poll trigger's recent polls, run Fetch now, and list live choices for a trigger's target fields, such as Slack channels or GitHub repositories.
— Names concrete MCP actions but lacks mechanism detail.4.0.0 (2026-11-09)
0700
Eval workflow management via MCPnew53how completely this was documenteddepth20/40specificity18/30actionability15/3053 / 100
Through the Runlayer Assistant or MCP clients, users can delete an eval, record a human verdict on an eval case, pick source runs from a plain-language description, label runs pass or fail, and group failed runs by cause.
Creating or updating an LLM gateway routing rule via the API or Terraform now returns a 422 error when the rule's target or any fallback is not a model listed on your AI providers. Add the model under Settings → AI models → Providers or use a model from `GET /api/v1/llm-gateway/models`. Existing rules continue routing; updates that leave targets unchanged still succeed.Our best guess
Microsandbox is an easy, fast, programmable and local-first microVM runtime and library
Microsandbox adds inbound TCPTCPA network protocol that delivers a reliable, ordered stream of bytes between two endpoints. It detects lost data and retransmits it. connection queuing and admission metrics for sandbox network traffic.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
1 FEATURE0 OF 1 FEATUREmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Inbound TCP queuing with admission metricsnew30how completely this was documenteddepth15/40specificity10/30actionability5/3030 / 100
microsandbox now queues inbound TCP connections and exposes admission metrics for sandbox network traffic.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
— No specifics on metric names or config to use themv0.7.14
Reality Defender detects synthetic audio, video, images, and documents for enterprise deepfake screening.
Reality Defender tightened scan timeout handling and free-tier enforcement with explicit error codes, improved audio analysis accuracy, and expanded YouTube URL ingestion to cover non-MP4 files and Shorts links.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
4 FEATURES2 OF 4 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0101
Stalled scan timeout reduced to 2 hoursbreaking90how completely this was documenteddepth30/40specificity30/30actionability30/3090 / 100
Scans stuck in ANALYZING now automatically end after 2 hours instead of the previous 12 hours, with status on GET /api/media/users/{requestId} set to UNABLE_TO_EVALUATE.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
Poll a scan and handle the new 2-hour timeout: check whether status is UNABLE_TO_EVALUATE and resubmit if so.
Free API plan blocks audio uploads with explicit errorbreaking75how completely this was documenteddepth25/40specificity30/30actionability20/3075 / 100
Audio uploads on the Free API plan now return HTTP 403 with code free-tier-restriction and message 'Forbidden: Video, audio, and text uploads require a paid plan', instead of failing silently as before.
YouTube URL ingestion improvementsimproved55how completely this was documenteddepth25/40specificity20/30actionability10/3055 / 100
YouTube URLs pointing to non-MP4 files are now downloaded and analyzed instead of failing, with failed downloads ending promptly rather than staying in DOWNLOADING; YouTube Shorts links submitted to the upload endpoint are now downloaded and analyzed as well.
Audio analysis now detects previously-missed speechimproved35how completely this was documenteddepth20/40specificity10/30actionability5/3035 / 100
Audio analysis now covers speech that was previously mistaken for silence and excluded from results; scores for existing files may differ as a result of this change.
Audio uploads on the Free API plan now return HTTP 403 (`free-tier-restriction`) where they previously may have succeeded — upgrade to a paid plan to restore access.Our best guess
Scans that remain in `ANALYZING` for 2 hours now resolve to `UNABLE_TO_EVALUATE`; integrations that expected up to 12 hours before timing out must be updated.Our best guess
Knostic enforces need-to-know access controls for enterprise AI assistants such as Microsoft Copilot.
Knostic's AgentMesh scanner added detection rules flagging two malicious 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. packages that secretly exfiltrate secrets at server launch.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
2 FEATURES1 OF 2 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0101
AgentMesh detection of mmdd-doc-mcp exfiltrationnew75how completely this was documenteddepth30/40specificity30/30actionability15/3075 / 100
AgentMesh now rates @ifasn/mmdd-doc-mcp@2.1.1 as Dangerous, detecting that it exfiltrates process command-line secrets over plain HTTP to 121[.]199[.]24[.]94 via the request path /beacon?stage=deepaudit&src=mmd-doc-mcp&res= on each server launch.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
AgentMesh detection of ms-doc-mcp environment harvestingnew75how completely this was documenteddepth30/40specificity30/30actionability15/3075 / 100
AgentMesh now rates @ifasn/ms-doc-mcp@2.1.1 as Dangerous, detecting that it records full values of environment variables matching KEY, TOKEN, SECRET, ALIBABA, DASHSCOPE, or CREDENTIAL to output/env_audit.txt and stderr as ENV-VALUE: lines on each server launch.
Arcade is the actions runtime for enterprise AI agents â one control point to enforce, execute, and govern every action, every time.
Arcade shipped a batch tool-execution history deletion command, paged Google Drive search, new GitHub, Linear, Slack, and Teams tools, per-execution API key and gateway reporting, and a breaking overhaul of file-conflict handling across its Office and Figma toolkits.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
└──▷ WHAT SHIPPED ·
11 FEATURES3 OF 11 FEATURESmost completely described firstwhat'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.
depth0–40what it does and how it works or what changed
specificity0–30names real surfaces — APIs, flags, formats, limits, numbers
actionability0–30enough 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.
0100
Linear toolkit tool and field enhancementsimproved80how completely this was documenteddepth30/40specificity30/30actionability20/3080 / 100
ListIssues can now filter results by creator (a specific person or the caller) and returns completion time, cancellation time, creator, and a sub-issue's parent; GetIssue returns completion and cancellation time; you can clear an issue's parent or project, or set a project's priority; ListWorkflowStates returns each workflow state's team key alongside its ID and name; and a new ListRecentComments tool lists every comment created or edited across a team or project since a given time, including quoted inline comments.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
File-conflict handling in Excel and Wordbreaking80how completely this was documenteddepth30/40specificity25/30actionability25/3080 / 100
Excel upload and create calls replace the previous plain overwrite flag with conflict_behavior (fail, rename, or replace); Word's create call now refuses a same-named file by default instead of silently replacing it, controlled by the same conflict_behavior argument. Existing calls using the old overwrite flag require migration.
Avoid accidental overwrites when creating a Word document by explicitly choosing what happens on a name conflict.
GitHub search and comment-listing toolsnew70how completely this was documenteddepth25/40specificity25/30actionability20/3070 / 100
New GitHub tools QueryIssues and QueryPullRequests search issues and pull requests across repositories using GitHub's search syntax, and ListRepositoryConversationComments lists every issue and pull request conversation comment in a repository in one paged, filterable and sortable call.
OpenFGA database creation controlimproved70how completely this was documenteddepth25/40specificity20/30actionability25/3070 / 100
Setting rbac.openfga.database.create: false makes the engine skip running createdb against the postgres maintenance database, letting operators point OpenFGA at a database a DBA already created with a least-privilege role.
Let OpenFGA use a pre-created, least-privilege database instead of trying to run createdb itself.
Batch tool execution history deletion commandnew60how completely this was documenteddepth30/40specificity15/30actionability15/3060 / 100
A new engine command deletes an organization's stored tool execution history on demand; it defaults to a dry run, requires a reason, and caps the number of rows removed per run.
Excel upload and create calls: the `overwrite` flag is replaced by `conflict_behavior` (`fail`, `rename`, `replace`) — existing calls using the old flag require migration.Our best guess
Word's create call now refuses a same-named file by default (instead of silently replacing it); the new `conflict_behavior` argument controls the behavior.Our best guess