The Cyber Toolchain — issue 068, September 8, 2026

Adds --compiler-scan / --no-compiler-scan flags to enable or disable a compiler-analyzer channel using gcc -fanalyzer or clang --analyze, plus --compiler-scan-max-tus <n> (default 2000) to cap the number of translation units analyzed.

$ /scan /path/to/c-project --compiler-scan --compiler-scan-max-tus 500 --build-command 'make -j4'

Adds --expanded-semgrep flag to re-run Semgrep rules over preprocessor-expanded views of macro-heavy C/C++ translation units, catching bugs that only appear in expanded output and that plain Semgrep misses.

$ /scan /path/to/c-project --expanded-semgrep --languages c,cpp

Adds --audit-verbose flag for strace-style verbose output during sandbox audit runs, and --audit-budget <n> flag to cap the maximum number of audit events collected per run.

$ /scan /path/to/target --audit --audit-verbose --audit-budget 500

The INVICTI_IGNORE_SSL_CHECKS environment variable in scan-cli now applies to all outbound platform API calls made during a scan — including DAST, Inventory, Identity, and report/export endpoints — not just the ephemeral agent.

$ export INVICTI_IGNORE_SSL_CHECKS=true
scan-cli scan --target-url https://app.internal.example.com

The INVICTI_MINIMUM_CONFIDENCE environment variable in scan-cli now independently fails a scan when its threshold is exceeded, without also requiring INVICTI_MINIMUM_SEVERITY to be set.

$ export INVICTI_MINIMUM_CONFIDENCE=confirmed
# INVICTI_MINIMUM_SEVERITY is intentionally not set
scan-cli scan --target-url https://app.example.com

Leaving the minimum severity field empty in the CI/CD script generator now produces a script with no minimum vulnerability threshold, applying to Azure Pipelines, CircleCI, 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., GitLab CI/CD, and Jenkins.

IDORIDORInsecure Direct Object Reference, a web vulnerability where an application exposes internal identifiers (IDs, filenames) without access checks, letting an attacker substitute another user's ID to reach their data./BOLABOLABroken Object Level Authorization, an API vulnerability class where a server fails to verify that the requesting user owns or may access the object ID they supply, letting attackers read or modify other users' data. authentication now supports Invicti Secrets Manager credentials (API key, basic auth, and 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. types), and the character limit for authentication values in IDOR/BOLA credentials has been increased from 1,024 to 4,096 characters.

⤢ ⤢ ⤢ ⤢

The Recon agent can now import architectural diagrams and documentation files in PDF, DOCX, and MD formats, enabling discovery of non-crawlable attack surfacesattack surfacesThe 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..

Assessment Profiles now support input caps and credit spend limits for more granular control over agent resource use; the OCTO sitemap now supports excluding URL paths and parameters, bringing it in line with DAST scan exclusion capabilities; and the DAST Engine now appears as a distinct entry under Attack Agents so total vulnerability counts align across AI agents and traditional DAST scans.

The scan engine now automatically throttles request pacing when it receives Retry-After or X-RateLimit-* rate-limiting response headers, reducing incomplete coverage and false-negative findings, and applies exponential backoff with jitter when retrying requests that receive a 429 Too Many Requests response with no Retry-After header.

Agentic Pentesting can now be managed directly from the AppSec Platform UI, with its findings flowing into the same pipeline as standard DAST scans for unified visibility and triage.

Bearer Token and API Key are now dedicated, first-class authentication header types in target configuration, separate from functional headers, for easier authenticated scan setup.

Two new request inspection settings, enabled by default: DATADOME_REMOVE_MATRIX_PARAMS strips matrix params (;) from URIs before regex evaluation, and DATADOME_ENCODED_CHAR_INCLUSION inspects static-looking paths containing encoded characters %23, %25, %2f, %3b, or %3f.

$ DATADOME_ENCODED_CHAR_INCLUSION=true
DATADOME_REMOVE_MATRIX_PARAMS=true

CloudFront v3 introduces DATADOME_STATIC_ASSET_EXCLUSION (enabled by default) to hold static-asset extension exclusions separately from custom URL exclusions, and moves custom URL exclusions to DATADOME_URI_REGEX_EXCLUSION exclusively. Static-asset extension patterns must be removed from DATADOME_URI_REGEX_EXCLUSION or they will duplicate or conflict with the new variable.

$ DATADOME_URI_REGEX_EXCLUSION='/health|/heartbeat'
DATADOME_STATIC_ASSET_EXCLUSION=true

The DataDome CloudFront Node.js module migrates from CommonJS (module.exports/require) to ES Module format (export/import), requiring .js files to be renamed to .mjs. The Lambda@Edge handler moves from a callback-based signature (exports.handler = (event, context, callback) => {}) to async/await (exports.handler = async (event, context) => {}), enabling compatibility with Node.js runtime 24 and later.

ggshield machine setup (requires ggshield 1.53.0+) configures AI Hooks for every detected AI agent, pre-commit/pre-push git hooks in the global git configuration, and a honeytoken decoy AWS credential profile in one idempotent run. Flags --no-honeytokens, --no-ai-hooks, and --no-git-hooks selectively skip individual protections. ggshield machine doctor reports per-AI-agent hook installation status, git hook status, and token scope validity. Planting honeytokens requires the honeytokens:write scope on ggshield auth login, gated behind the workspace Manager role, while the install script provisions default honeytokens:check, endpoints:send, and ai-discover:send scopes.

$ ggshield machine setup

$ ggshield auth logout
ggshield auth login --scopes honeytokens:write
ggshield machine setup

$ ggshield machine setup --no-honeytokens

$ ggshield machine doctor

$ ggshield machine setup

Blocks secrets at prompt submission, pre-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., and post-tool use across AI coding agents, now including Codex and Mistral Vibe alongside existing Claude Code, Cursor, Copilot CLI, and VS Code coverage; blocks surface a named secret and remediation hint. Self-hosted instances require ClickHouse to support machine scans with secrets findings, MCP tool calls, and full AI agent session recording.

Adds detector, detector type, secret family, secret category, and secret provider filters and sort options to the NHI inventory. Groups multi-match credentials (e.g. PostgreSQL username, password, host, and port) as a single inventory entry rather than separate fields, and labels unmatched secrets as Unknown so they remain visible and filterable rather than dropped.

ggshield ai discover --history parses MCP tool calls from agent transcripts to reconstruct which tools an agent called, for blast-radius analysis after a suspected credential compromise.

$ ggshield ai discover --history

$ ggshield ai discover --history

Adds new Firebase ID Token, Coze Access Token, Keycloak Access Token, Keycloak Refresh Token, Fish Audio API Key, and Azure Custom Vision Training Key detectors, a new Apify API tokens analyzer, and docker login --username/-u detection to the Docker Credentials detector for Docker Hub personal access tokenspersonal access tokensA personal access token, a credential used by scripts and integrations instead of a user's password. Its permissions and expiry depend on the service issuing it..

Runs the Secrets Detection Engine on every value ggscout collects, classifying vaulted secrets against 550+ detectors including multi-match detectors (ggshield 1.54.0, ggscout 0.32.0).

Extends New Relic Key and New Relic Insights Insert Key checkers to identify revoked keys via an empty POST request. Improves scaleway_token precision by rejecting node UUIDsUUIDsA 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. from proxy-subscription configs, and improves fastly_personal_token precision by rejecting proxy-subscription node passwords and 32-char prefixes of longer base64url values.

Adds AI agents and MCP inventory to Developer Endpoint Protection, reporting agents and MCP servers per endpoint, hook coverage, and whether an agent subscription is personal or company-owned.

AWS access keys stored in a secret manager or Kubernetes now link to their 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. identity by AccessKeyId, showing the permissions of the stored key in the NHI GovernanceNHI GovernanceThe policies, controls, and lifecycle management applied to Non-Human Identities — service accounts, API keys, tokens, and certificates — to enforce least-privilege access and reduce credential sprawl across systems. exploration map.

Adds debug configuration option in veracode.yml for debugging and troubleshooting pipeline runs (GitHub Workflow Integration v1.0.6, Azure DevOps Workflow Integration v2.3.6); adds analysis_branch support so scheduled scans run against a specified branch instead of the default branch (Azure DevOps Workflow Integration v2.3.10); adds a custom agent pool name parameter to specify the agent pool used for pipeline execution (Azure DevOps Workflow Integration v2.4.3); and adds support for custom predependency commands to prepare projects before scanning (GitHub Workflow Integration v2.1.3).

# In veracode.yml
debug: true

# In veracode.yml
analysis_branch: release/2026-q3

Adds page size support to the Platform Applications API (/appsec/v1/applications) to improve efficiency retrieving large application lists (v26.3.19.0); adds support for the Veracode Platform Findings 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. to access open and mitigated application findings (v26.2.18.0); introduces support for the paginated Sandbox REST API enabling retrieval of sandbox lists using pagination (v26.3.19.0); and implements a configurable retry mechanism for the Applications API (appsec/v1/applications) to improve handling of transient failures and timeouts (v26.2.18.0).

The Veracode Healthcheck endpoint /healthcheck/status has been removed and now returns a 404 error; use the v2/principal endpoint in the Identity REST API instead.

The Veracode Flaw Importer task (containing spaces in the task name) is deprecated and replaced by VeracodeFlawImporter in the Veracode Azure DevOps extension v3.35.0; the deprecated task is scheduled for removal in a future release.

Adds the most_recent_scan filter to the Reporting API Scans report, which when set to true returns only the most recent scan per asset; adds the custom_severity filter to the Findings report; introduces a new Audit Log report with an action_type filter to query user activity from Veracode Analytics; adds the collections_list field to the Finding Reporting API response listing collection names for an application; adds the library_instance_ref field to uniquely identify the library instance tied to a finding; and adds the data_refresh_date field indicating the most recent Analytics data refresh time.

Support for the Veracode Integration for Jira on Jira Server and Jira Data Center 9.x ends November 30, 2026 — organizations must upgrade to Jira 10 or later to maintain a supported configuration.

Adds support for OAuth Client Credentials authentication with the Veracode XML APIsXML APIsInterfaces for reading, writing, or querying XML documents, provided by libraries such as libxml2 or built into language runtimes; cyber tools use them to parse structured data like configs, reports, or protocol messages reliably., previously limited to REST APIsREST APIsA convention for building HTTP-based interfaces where resources are addressed by URL and manipulated with standard verbs like GET, POST, and DELETE, giving cyber tools a predictable way to integrate with external services., and adds the ability to create and manage 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. Client credentials on the API Credentials page in the Veracode Platform.

Makes container and IaC scan results performed from the CLI available on the Veracode Platform, with severity counts, filtering, scan history, and mitigation guidance.

Introduces Manual Penetration Testing (MPT) Flaw Matching to correlate findings across tests for consistent flaw IDs, status values, and remediation grace period support.

Adds a Current File filter to Static Scan Findings in VS Code to view findings for the file currently open in the editor, plus an option to filter and group findings by file name, with findings for each file grouped by severity (Veracode Scan for VS Code v1.16.3).

Enables scheduling Static Analysis, 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., and IaC scans for GitHub organizations directly from the Veracode Platform (GitHub Workflow Integration v2.0.0), and enables scheduling scans for every organization where the Azure DevOps Workflow Integration is installed, based on veracode.yml configuration (Azure DevOps Workflow Integration v2.3.9).

Enables installing and configuring the Veracode GitHub Integration directly from the Veracode Platform, eliminating manual GitHub Marketplace installation steps (GitHub Workflow Integration v2.0.0), and adds a new uninstallation workflow on the Platform to remove the Veracode app from GitHub organizations (GitHub Workflow Integration v2.1.1).

Updates the Auto-Update OWASPOWASPA nonprofit foundation that publishes open standards, guides, and ranked vulnerability lists—most famously the OWASP Top 10—giving cyber tools a vendor-neutral benchmark to test or report against. rule and Auto-Update CWECWECommon Weakness Enumeration, a catalog of recurring software and hardware weakness types. It describes classes of mistakes, such as injection, rather than individual vulnerability instances. Top 25 rule in application security policies to enforce 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. and MITREMITREA US federally funded research and development center that maintains public frameworks—ATT&CK, CVE, D3FEND—that cyber tools use to classify threats, vulnerabilities, and defensive techniques in a common language. CWE Top 25 for 2025.

Introduces a new IDE and CLI Usage Explore and Dashboard in Veracode Analytics to visualize organization-wide IDE and CLI plugin usage; adds a Firewall Usage Explore and Dashboard surfacing metrics for Veracode Firewall; and adds Collections fields to the Applications explore, including new Collections visualizations in the Policy Compliance Overview dashboard.

Mend Platform 26.8.2 introduces a unified vulnerability remediation experience with a Top Fix column in the Findings table, an updated Remediation tab, and standardized fix logic and metadata across the Mend Platform. The updated fix logic may result in fewer remediation suggestions per vulnerability than before, so workflows relying on the previous suggestion volume will see reduced recommendations.

The JavaScript/TypeScript analysis engine's detection accuracy was improved across 11 CWEs: CWE-22 (Path Traversal), CWE-78 (OS Command Injection), CWE-79 (XSSXSSCross-site scripting, an attack that runs malicious script in another user's browser through a trusted site. It can steal data or perform actions using the user's session.), CWE-89 (SQL Injection), CWE-90 (LDAP InjectionLDAP InjectionAn attack technique where unsanitized user input is embedded into LDAP queries, manipulating directory lookups to bypass authentication or extract data. Tools use it to test directory services for input-validation flaws.), CWE-94 (Code Injection), CWE-117 (Log Neutralization), CWE-295 (Certificate Validation), CWE-601 (Open Redirect), CWE-918 (SSRFSSRFServer-side request forgery, an attack that tricks a server into making requests chosen by an attacker. Those requests can reach internal services the attacker cannot access directly.), and CWE-943 (NoSQL/query-language injection), including tuned handling of React controlled components to reduce false-positive DOMDOMThe browser's in-memory tree representation of a web page's structure, standardized by the W3C, giving cyber tools a programmable interface to read or manipulate page content at runtime.-based XSS findings.

A new suppression tool in the IDE and AI agent workflow supports single and bulk suppression or unsuppression of findings across all scanning engines, including SAST and 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., with UI updates reflecting within seconds.

The Developer Platform now supports scanning multiple base branches into dedicated projects on Azure DevOps and Bitbucket Cloud, enabling security managers to monitor and set policies for each branch individually.

StepSecurity now supports up to ten webhooks per tenant, each with its own endpoint URL, HTTP method, headers, message format, and Envelope template — replacing the previous single-webhook-per-tenant limit. Each webhook can subscribe to any combination of four event types (Insights, Detections, Secure Registry Audit Logs, API Audit Logs) and can be scoped to a subset of repositories or to all current and future repositories, with independent delivery per endpoint so a failure at one does not affect others. The UI adds an Add Webhook control that creates additional endpoints, each shown as a collapsible card with name, endpoint, and event-type chips.

⤢ ⤢

Improves detection accuracy of the JavaScript/TypeScript analysis engine across CWECWECommon Weakness Enumeration, a catalog of recurring software and hardware weakness types. It describes classes of mistakes, such as injection, rather than individual vulnerability instances.-22, CWE-78, CWE-79, CWE-89, CWE-90, CWE-94, CWE-117, CWE-295, CWE-601, CWE-918 and CWE-943, including tuning for React controlled components to reduce false positives where query parameters flowing into escaped JSX attributes were incorrectly flagged as DOMDOMThe browser's in-memory tree representation of a web page's structure, standardized by the W3C, giving cyber tools a programmable interface to read or manipulate page content at runtime.-based XSSXSSCross-site scripting, an attack that runs malicious script in another user's browser through a trusted site. It can steal data or perform actions using the user's session..

Renovate now passes --no-frozen-lockfile to pnpm when performing lock file updates in the manager/npm manager, supports optional channel options in the manager/pixi manager, and supports AWS S3 as a Helm charts repository in the datasource/helm datasource.

Adds a Top Fix column to the Findings table and an updated Remediation tab, standardizing fix logic and recommendation metadata across SAST and 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. findings on the Mend Platform. Because the new fix logic is more accurate, it may surface fewer remediation suggestions per vulnerability than before, reducing noise but potentially removing previously shown suggestions — existing workflows relying on the old suggestion volume should be reviewed.

Adds a suppression tool in Agentic Integrations supporting single and bulk suppression or unsuppression actions across all scanning engines, including SAST and SCA, with UI updates reflecting within seconds to reduce context-switching for developers working in their IDE or AI agent workflow.

Mend SAST now supports scanning multiple base branches into dedicated projects in the Developer Platform on Azure DevOps and Bitbucket Cloud, letting security managers monitor and set policies for each branch individually.

Adds scoped permissions controlling Rules, repository context, and scan locations that surface only the actions a member or team is authorized to perform, with inline reasons when Add, Save, or Delete are unavailable. Editing or deleting a rule, repository context, or scan location set is now authorized against both the scope a definition already holds and the scope being assigned, blocking moves outside a member's granted repositories. The teams permission editor lists every grantable permission, and an organization-wide repository permission grant carries its named actions — including saving repository context or scan location sets — across all repositories; members with per-repository scanner permission see a searchable repository list under Scanner scoped to repos they can configure, and definitions linked to a repository removed from the version control host remain deletable.

Adds a dedicated Scanner workspace in the main sidebar that consolidates all scanning configuration under Configuration (scanner settings, scan locations) and Detection (repository context, rules, sources and sinks). Settings sidebar pages are now grouped by purpose — Workspace (general, members, teams, billing), Security (scanner settings, repository tags, service levels), and Connections (integrations, notifications, API keys) — and scanner settings appear across five consistent categories, General, Full scans, Pull requests, Dependencies, and Remediation, at organization, tag, and repository scope so overrides appear in the same place at every level.

Adds Settings → Service levels → Policies to configure SLO targets on individual policies, a Settings → Service levels → SLO / SLI section with trend charts turning SLASLAA formal agreement between a service provider and customer that defines expected uptime, response times, and remedies for failures, giving cyber tools a contractual baseline for availability and support commitments. activity into a time-series view, and an Audit trail section under Settings → Service levels to record SLA-related activity.

Adds a Hide Descriptions toggle in the Repositories & data section for organization admins, which redacts AI-generated descriptions and exploit walkthroughs for secret findings across the dashboard, API responses, webhook payloads, notifications, and exported tickets.

Adds a --oci-store-user flag to sudo ig run so gadgets can be read from a per-user OCI store at ~/.ig/oci-store (built rootlessly) without creating root-owned files in the user home directory, and enables ig image build, ig image list, ig image inspect, ig image push, and ig image pull to run as a normal non-root user against that same store. This is a breaking change: gadgets previously built into the root-owned store will not be visible to rootless ig image commands.

The uprobetracer now falls back to Go's .gopclntab section (including the Go 1.26+ .go.module format) when the ELFELFThe standard binary format for executables, shared libraries, and object files on Linux and most Unix-like systems, giving cyber tools a predictable structure to parse, instrument, or analyze compiled code. symbol table is absent, enabling uprobes on stripped Go binaries such as a production kubelet.

Container runtime socket auto-detection now uses the kubelet's /configz endpoint over HTTPSHTTPSHTTP carried over an encrypted TLS connection. It protects traffic in transit and lets a browser verify the server's certificate., requiring only the nodes/configz RBAC permission instead of the broader nodes/proxy (Kubernetes v1.33+; earlier clusters fall back to the configured socket path). This is a breaking change: v1.33+ deployments must update RBAC rules to grant nodes/configz instead of nodes/proxy.

The Kubernetes collector manager is now published as a multi-architecture image supporting both amd64 and arm64, enabling deployment on Graviton node pools without requiring a dedicated x86 node or nodeSelector configuration.

Enterprise Plus and Google Unified Security (GUS) customers can now download the Bindplane Enterprise (Google Edition) license key directly from the platform console, under SIEM Settings > Collection Agents > Bindplane Enterprise (Google Edition), to activate a self-managed deployment.

New v2 feed types use the Google Cloud Storage Transfer Service (STSSTSAWS Security Token Service, an Amazon Web Services API that issues short-lived credentials for IAM roles, letting cyber tools authenticate to AWS without storing long-term secrets.) for improved performance, scalability, and reliability. Feeds using v1 feed types will be automatically migrated to v2 starting April 6, 2026, and some feeds may require additional IP allowlist or service account permission updates before that migration.

Auto CDF computes row-level changes at query time using row tracking, eliminating the need to manually run ENABLE CHANGE DATA FEED on each Delta Lake or Apache IcebergApache IcebergAn open table format for large analytic datasets, maintained by the Apache Software Foundation, that gives cyber tools reliable schema evolution, time-travel queries, and atomic commits on object storage or HDFS. v3 table. It makes MERGE and UPDATE operations approximately 15% faster on tables queried for changes by removing write-time overhead, and works with batch queries, Structured Streaming, and Delta Lake Sharing when using Databricks Runtime 19 LTS or above with row tracking enabled.

Quick Filters and Advanced Filters are merged into a single Filters experience in the dashboard Filters drawer, grouped by Source (Organizations, Sites, Roaming Clients/Relays, Users) and Traffic (Result).

Dashboard deployment widgets (Roaming Clients Protected, Sites Protected, Relays, Users, Collections) now show a pulsating green dot with a tooltip indicating the data is live and not affected by the time range filter.

The FIREZONE_NAME environment variable and -n CLI flag have been removed from the Gateway; the Gateway's name must now be configured in the Admin Portal. This is a breaking change for any deployment scripts or container configs that set FIREZONE_NAME or pass -n.

Advanced MFA Configurations is now GA, enabling configurable Remember this device idle lifetimes (1 hour to 30 days) and absolute lifetimes (up to 90 days), 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. lengths for SMSSMSShort Message Service, the protocol built into cellular networks for sending text messages between phones; cyber tools use it for alerting, two-factor authentication delivery, or phishing simulation over mobile channels. and email factors (4–10 digits), OTP lifetime for SMS MFA (30 seconds to 1 hour), and OTP lifetime for email MFA (5 minutes to 1 hour). Configured via Dashboard under Security > Multi-factor Auth by setting OTP Length and OTP Lifetime per factor.

Auth0 now supports acting as a requesting application in Cross-App Access flows (Early Access), built on Auth0 Token Vault using the ID-JAG (Identity Assertion Authorization Grant) protocol, letting apps fetch third-party API tokens for enterprise users without a manual Connect your account step. Supported on 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 Okta Workforce connections with org-aware resolution via org_id; a new Dashboard toggle enables Cross-App Access Requesting App on enterprise connections.

⤢ ⤢

Adds self-service configuration of Third-Party Application Access and Cross App Access (XAA) through the My Organization API and Embeddable UI Components (Open Early Access), letting Organization Admins configure third-party app access, connection-level third-party access, and XAA Resource Application roles without routing changes through the Auth0 Dashboard.

ESV secrets are now stored in Google Secret Manager using a unique name-and-version identifier as the default JWK kid; deployments where multiple secrets shared the same GSM version must set the new esv.am.secrets.gsm.stableid.version.only environment variable (esv.am.secrets.gsm.stableid.version.only=true) to retain the legacy kid value.

$ esv.am.secrets.gsm.stableid.version.only=true

Adds IGA role composition certification, enabling reviewers to certify role definitions including attributes, glossary data, entitlements, applications, and membership rules, with a dedicated review task view for role-composition items.