This Week's Highlights — issue 021, November 20, 2026
The best tooling updates that shipped this week.
The week in numbers · Nov 13–19
Changes per week, last 12 weeks · Nov 13–19
Changes we read in that week's dailies, including docs and API changes. A gap means no daily was published.
Category moves this month
Every feature release from the 1326 tools on our watchlist goes into the daily newsletter. Once a week we read the whole field side by side and grade each release from 1 to 5 on three things, weighted equally:
- Depth. How complete is the best new capability?
- Impact. How much does it change what you can do?
- Novelty. Has any other tool done this before? We check every capability against the history of every tool we track. Features most tools already have earn nothing here. Being first earns the most.
A tool is judged on everything it shipped that week, so a project that releases daily gets credit for the sum — and still only takes one slot. We would rather run a short list than a padded one.
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.
This issue came out before editions were ranked, so it shows the Standard edition in issue order.
The picks in full
Humanix detects and stops social engineering attacks as they unfold across email, chat, voice, and help desk workflows.
Humanix adds Complete Call Context, spawning a micro-agent per sensitive service-desk request that correlates live call behaviour with IdP and ITSM records in real time — flagging mismatched managers, non-existent ticket numbers, or unverified identity claims before the call ends, closing a gap no other tracked tool addresses.
Humanix adds Complete Call Context: per-request micro-agents correlate live service-desk calls with IdPIdPA service that authenticates users and supplies identity information to other applications. It is the sign-in authority behind many single sign-on systems. and ITSMITSMIT Service Management: a framework of practices for delivering and supporting IT services, covering ticketing, change control, and incident response. Cyber tools integrate with it to automate alert-to-ticket workflows. records to detect vishing in real time.
- New detectionNew integrationAdds Complete Call Context, which spawns a micro-agent per sensitive service-desk request (such as an MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. reset or password change) to correlate call behaviour with 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. and IT service management records, then raises a severity-rated alert with a plain-language explanation when any discrepancy or missing evidence is found. Not checked yet
- Micro-agents now check caller claims against directory, ticketing, and identity provider records in real time — flagging conflicts such as a mismatched manager, a non-existent ticket number, or an unverified identity claim — rather than waiting until after the call. Not checked yet
2 details in Everything
Zenity discovers, secures and governs enterprise AI agents from buildtime to runtime across SaaS, endpoint and cloud.
Zenity now treats agent sandboxes as a first-class security surface: continuous discovery of every sandbox and template, posture checks that flag real API keys or wildcard egress, active red-team testing inside each sandbox profile, and runtime enforcement that evaluates actions against the full session context — no other tracked tool covers this attack surface.
Zenity adds a full sandbox security layer: discovery, posture checks with active red-team testing, runtime protection with taint and intent tracking, four containment levels, a trajectory view, and a new Sandboxes policy page.
- New integrationA new Sandboxes page lets security teams write per-agent file, network, and runtime policies — including specific blocked paths, egress allow-lists, and environment variable listings by name — that the sandbox enforces directly. Not checked yet
- New sandbox discovery continuously inventories every sandbox and sandbox template, linking each to its agent, credentials, workspace, egress policy, owner, image, and mid-session-installed packages, skills, 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. servers, including shadow sandboxes and sandboxes without a managed platform. Not checked yet
- Posture checks now flag misconfigured sandboxes — such as real API keys in the environment or image, wildcard egress rules, internet-reachable sandboxes, or root-running containers — as posture violations with risk descriptions and remediation guidance. Not checked yet
- Posture is now also tested actively: a red-team agent runs inside each sandbox profile attempting to escape, reach denied hosts, read credentials, or tamper with hooks, and every path found becomes a hardening rule for that profile. Not checked yet
- New detectionNew integrationUpdated integrationRuntime protection now evaluates each action against the full session context — combining harness intent, taint from untrusted content, and low-level process, file, and network telemetry — blocking actions that conflict with the agent's declared task. Not checked yet
+6 more show less
- Zenity Boundaries now extend enforcement into the sandbox itself, applying the same rules used in the agent harnessagent harnessThe software around an AI agent that manages its tools, execution loop, state and limits. It determines how the model's proposed actions are run and checked. across any supported platform. Not checked yet
- Four containment levels — Steer, Restrict, Contain, and Kill-and-recover — are now available directly from a finding, or can be triggered automatically by a boundary, with each step preserving evidence and rotating exposed credentials. Not checked yet
- A trajectory view now shows every network request, process, and file action on a single timeline alongside the prompt and tool call that caused it, spanning from intent to kernel. Not checked yet
- On supported platforms, posture checks now extend to saved snapshots, verifying them before an agent resumes or forks. Not checked yet
- AI Security Posture Management (AISPM) now gains sandboxes and sandbox templates as asset types, linked to their workspaces, credentials, and policies. Not checked yet
- AI Detection and Response now gains visibility into sandbox-level activity beyond what the agent itself reports, with the full sandbox history available for investigation. Not checked yet
11 details in Everything
- New integration Sandbox platform integrations
- New supported platform Vendor-operated cloud coding agents
- New integration Claude Enterprise in Microsoft 365 — inline prevention, telemetry, and findings
- New integration Microsoft Copilot Cowork — telemetry and findings via Cowork transcripts
- Updated integration Microsoft Sentinel integration upgraded from raw-log export to high-fidelity alerts
1 more release this week · 2026-11-18 → 2026-11-19
Wallarm ships seven autonomous Agentic Operators — Radar, Sift, Trim, Tune, Lens, Bard, and Wally — that run the full WAAP operations loop (attack-spike investigation, hourly false-positive clearing, rule-set pruning, policy tuning, inventory review, posture briefing, and on-demand configuration) without prior art among tracked tools.
Wallarm introduced seven AI Agentic Operators that autonomously triage alerts, prune rules, and tune policies, manageable via 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 and an audit-logged autonomous mode.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
Wallarm added seven named Agentic Operators: Radar, which detects attack spikes, investigates them, and suggests a response; Sift, which clears false positives on an hourly cycle so alerts remain actionable; Trim, which tidies the active ruleset to prevent rule sprawl; Tune, which adjusts protection policies to match observed application behaviour; Lens, which reviews the API inventory and flags risky endpoints; Bard, which converts the current security posture into a human-readable briefing; and Wally, which answers questions and handles configuration requests on demand.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
All Agentic Operators can be managed through Wallarm's MCP server, letting them plug into external agentic workflows without returning to the console; a customer success team enables Agentic Operators, after which an agentic workflow connects to the MCP server to invoke operators such as Sift, Radar, and Wally programmatically.
Agentic Operators support an autonomous mode where routine approved actions run without manual confirmation, with every finding logged to the audit trail.
Auth0 is an identity and access management platform that provides authentication, authorization, and user management services for applications via APIs and SDKs.
Auth0's Security Agent detects credential stuffing attacks that stay under rate limits by computing per-tenant traffic baselines and alerting when failed logins deviate from normal for a sustained period — a class of attack that threshold-based controls miss entirely — then triages the attack and recommends specific Attack Protection and Tenant ACL configuration changes an admin can apply in one step.
Auth0 shipped three distinct capabilities this window: general availability of 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. for third-party apps, a beta in-app sign-in flow called Embedded Authorize, and a reworked Email Templates editor with live preview and agent-prompt handoff.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
A new POST /e/authorize endpoint lets apps build in-app sign-in without redirecting to Universal Login; it returns the allowed next steps for the app to render until Auth0 issues an authorization code.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ During Beta it supports email and phone one-time-code sign-in and second-factor MFAMFAMulti-factor authentication, requiring proof from different categories, such as a password and a hardware key, to sign in. It reduces reliance on a password alone. (email OTPOTPOne-time password: a short-lived numeric or alphanumeric code valid for a single authentication attempt, generated via TOTP or HOTP standards, giving cyber tools phishing-resistant second-factor verification without stored shared secrets., phone OTP, TOTPTOTPA one-time password algorithm (RFC 6238) that generates short-lived codes from a shared secret and the current time, giving cyber tools phishing-resistant second-factor authentication without a server round-trip., push, recovery code), and works out-of-the-box with Auth0 Universal Components for web, iOS, and Android.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
$ curl -X POST https://<your-tenant>.auth0.com/e/authorize \ -H 'Content-Type: application/json' \ -d '{"client_id": "<client_id>", "redirect_uri": "<redirect_uri>", "response_type": "code", "scope": "openid profile email"}'
Third-party apps can now use OpenIDOpenIDAn open standard for decentralized authentication that lets a user prove their identity to a site via a trusted third-party provider, so tools can delegate login without managing credentials themselves. Connect, with per-app control over which standard scopes (profile, email, address, phone) they may request via OpenID Connect Settings in the dashboard. Dynamically registered apps (via Client ID Metadata DocumentsClient ID Metadata DocumentsOAuth 2.0 documents hosted at a client's own URL that declare its identity and redirect URIs, letting authorization servers fetch and verify client metadata without manual registration.) can't receive 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. settings directly, so OIDC is instead configured through Application Profiles, which all apps registered under that profile inherit; third-party apps now get profile claims in the ID token while the /userinfo endpoint returns only sub for them.
A dedicated editor at Emails > Templates in the dashboard lists all templates with status and opens each on its own detail page, where editor and live preview tabs show the From line, Subject line, and rendered email. A new Variables panel lets users search and copy template variables, and a Copy agent prompt action copies the template's variables, current body, and deploy steps for handoff to a coding agent.
- New detection Credential stuffing attack detection
StepSecurity ships two first-of-kind supply chain controls this week: Secure Registry routes npm/pnpm/Bun installs through a cooldown proxy that blocks known-compromised packages and holds new versions for a configurable window, integrated into Harden-Runner with a single flag; and Harden-Runner now enforces network, DNS, process, and HTTPS monitoring inside Kata Containers and Amazon EKS Nitro Sandbox microVMs, closing a gap where CI hardening stopped at the VM boundary.
StepSecurity's biggest release this window is a free community tier for its Secure Registry, which intercepts npm/pnpm/Bun installs in Harden-Runner to block compromised packages and enforce a cooldown; it also shipped a new fork-PR approval audit control and remediation tracking for Threat Center.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄
what's the number?
Each feature carries 0–100 for how completely the vendor documented it — not how big or important the work is. A major capability described in eight words scores low, and that is the finding.
- depth0–40 what it does and how it works or what changed
- specificity0–30 names real surfaces — APIs, flags, formats, limits, numbers
- actionability0–30 enough to go use it — a named endpoint, flag, or config key tops this; a UI path is a starting point
Code and config rank first by construction: a documented endpoint, flag or config key scores at the top of specificity and actionability, so it sorts above a feature described only in prose. Under 60 the notes go thin — everything below the “thinner coverage” line is thinner documentation, not smaller work. Hover any meter for that feature's three sub-scores.
Enable protection for any GitHub ActionsGitHub ActionsGitHub's built-in CI/CD system that runs automated workflows triggered by repository events, letting cyber tools automate testing, building, and releasing without managing separate pipeline infrastructure. job by adding secure-registry: true to a step-security/harden-runner@v2 step, which routes npm, pnpm, and Bun 1.4.4+ installs through community.registry.stepsecurity.io/javascript/ and blocks known-compromised packages via StepSecurity 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. flagging (always on).︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Installs are held back for a configurable cooldown of 0–30 days (default 2 days) using the secure-registry block's npm.cooldown.days key, with exemptions for specific packages/versions such as typescript@* or [email protected]; a custom block list can also refuse specific packages or versions.︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄ Disable the registry for a job (e.g. one running npm publish) with npm: mode: off, and inspect held/blocked versions in the job summary (with a ready-to-paste allow snippet) or review every request on the new Packages tab in Harden-Runner Insights, which shows request type, status, and evaluated controls per download.
- name: Harden the runner
uses: step-security/harden-runner@v2
with:
egress-policy: audit
secure-registry: true
- name: Harden the runner
uses: step-security/harden-runner@v2
with:
egress-policy: audit
secure-registry: |
npm:
cooldown:
days: 3
exemptions:
- "typescript@*"
- "@typescript/*"
npm publish can reach the public registry directly without being intercepted.- name: Harden the runner
uses: step-security/harden-runner@v2
with:
egress-policy: audit
secure-registry: |
npm:
mode: off
- name: Harden Runner
uses: step-security/harden-runner@v2.23.0
with:
secure-registry: true
Threat Center findings can now be marked remediated individually (via row menu, with an optional comment) or all at once in the current view via Remediate all; the incident API endpoint now returns each finding's remediation status, and new endpoints let you mark findings remediated or reopen them for one organization or tenant-wide. Matched components and Matched IOCsIOCsAn indicator of compromise, an observable clue that may point to malicious activity, such as a suspicious file hash or domain. It is evidence to investigate, rather than proof by itself. tables show status, who remediated each finding, when, and their comment, with All, Active, and Remediated filter tabs plus source-based search; incidents move to a Remediated tab once fully cleared and move back automatically if a finding reappears, flagged with a 'Reappeared:' title prefix and a notification through existing threat intel channels. A summary banner shows outstanding vs. fully remediated findings, every action is recorded in audit logs, and remediating requires Read & write access to the Threat Center permission.
A new 'Fork pull request workflows should require approval' control, rated Critical and listed under Code Injection in the Overview dashboard, checks every GitHub organization and public repository's fork PR workflow approval policy every hour and links directly to the fix in GitHub's Settings > Actions > General. Findings are also surfaced through the existing control tools in the StepSecurity 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, and the check requires the StepSecurity GitHub App to have Administration: read permission on the organizations and repositories it scans.
- New supported platform Conda package inventory and configuration audits
- New supported platform Java package inventory and configuration audits
- New supported platform Swift package inventory and configuration audits
- New supported platform Pi coding-agent package and extension inventory
- New supported platform Support for pooled and multi-session Windows virtual desktops
Where each pick stands · Nov 13–19
Tools with changes in each pick's category over the last 12 weeks, including the pick itself.
Each pick's release rhythm · Nov 13–19
Changes per week for the last 12 weeks. The bright bar is this week; a dash marks a week with no daily.
Which categories moved · Nov 13–19
The thick bar is this week. The thin bar is the average of the four weeks before, counting only weeks with a daily.
Who shipped the most · Nov 13–19
Changes per tool this week. Volume is not a quality score.