StepSecurity

No entries match these filters. Clear filters

1. GitHub Checks: Pass on Non-Public Repositories for PWN Request and Script Injection (2026-11-03) Nov 3, 2026 · issue 130

   StepSecurity adds a Pass on non-public repositories setting for PWN Request and Script Injection controls in GitHub Checks⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢
   └──▷ HOW TO FIND IT
   Reduce alert noise from private and internal repos while keeping strict enforcement on public repos for PWN Request findings.
   📍In the StepSecurity console, go to GitHub Checks › Configuration, select the PWN Request control, and enable Pass on non-public repositories.
2. Malicious Files: Suppression Rules and Finding Actions (2026-11-03) Nov 3, 2026 · issue 130

   StepSecurity adds malicious-file suppression rules and finding actions for Dev Machine Guard and GitHub⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢
   └──▷ HOW TO FIND IT
   Suppress a known test-fixture binary fleet-wide before it floods every future scan with false positives.
   📍In Dev Machine Guard, go to Suppression Rules › New Rule. Set File path to the fixture path (e.g. tests/fixtures/malware_sample.exe), set Detection rule to the matching rule name, leave Device as All devices, choose reason Used in tests, then click Preview matching findings to verify scope before saving.
   Quickly scope a suppression to exactly one endpoint when triaging a finding that is unique to a developer's machine.
   📍On the Dev Machine Guard Findings page, open the finding, click Create suppression rule (file path and detection rule are pre-filled), then click Limit to this device, select reason Accepted risk, add an optional comment, and save.
   Suppress a reviewed test fixture in GitHub so every existing and future version of the file is silenced without repeated manual triage.
   📍In StepSecurity GitHub Findings, locate the file finding, click Suppress, select reason Used in tests, add an optional comment, and confirm — the status applies to all versions of the file across every branch, pull request, and tag.
   • Dev Machine Guard now has a Suppression Rules page where you define fleet-wide rules that match on file path, detection rule, SHA-256SHA-256A cryptographic hash function that turns data into a fixed 256-bit fingerprint. It is used to check integrity; hashing does not encrypt the original data. hash, campaign, confidence, or any combination — with * as a wildcard — and apply to all devices or a selected subset.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Suppression rules apply to current findings and to files found on later scans, and you can preview matching findings before creating a rule, then edit or delete rules at any time.
   • Create a suppression rule directly from a finding, with its detection rule and file path pre-filled, and optionally limit the rule to that finding's device in one click.
   • Resolve or suppress individual findings on Dev Machine Guard with an optional comment; suppressed findings display their reason in a new Reason column.
   • Deactivate a detection rule behind a finding for every device without leaving the Findings page, with a required reason for the change.

   +1 more show less

   • Suppressed findings on GitHub now show their reason in a new Reason column; new file versions inherit the suppressed or resolved status until explicitly changed.

   6 details in Everything

3. Harden Runner Releases: October 29, 2026 (2026-10-29) Oct 29, 2026 · issue 125

   StepSecurity Harden Runner adds Avrea-hosted runner support, a flag to block implicit AWS endpoint allowances, and configurable 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. monitoring buffers with PrometheusPrometheusAn open-source monitoring system maintained by the CNCF that scrapes and stores time-series metrics, giving cyber tools a standard way to expose and query operational data. health⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢
   └──▷ USE IT
   Prevent the agent from silently opening AWS endpoints on a self-hosted Linux runner — useful when your policy requires all allowed endpoints to be explicitly declared.
   yaml copy

   env:
     disable_implicit_endpoints: true

   Tune buffer sizes on a high-throughput ARC node where DNS lookups or process events are being dropped, and confirm the sizing is working via Prometheus metrics.
   yaml copy

   env:
     dnsTrackingMapSize: 65536
     processRingBufferSize: 32768
     enableMetrics: true

   • Disable implicit AWS endpoint allowances on self-hosted Linux runners with the disable_implicit_endpoints feature flag in Self-Hosted VM Harden Runner v1.9.5; GitHub and StepSecurity endpoints remain allowed.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Report eBPF monitoring health metrics — ring buffer usage, dropped network/DNS/process events, and DNS tracking map inserts, removes, and errors — via the Prometheus endpoint when env.enableMetrics is enabled in ARC Harden Runner v2.24.0.
   • Harden Runner now supports runners hosted by Avrea (enterprise tier) in GitHub Action v2.22.3.
   • A new feature flag in Harden-Runner GitHub Action v2.22.3 stops implicit AWS endpoint allowances on self-hosted Linux runners (enterprise tier), while GitHub and StepSecurity endpoints remain allowed.

   4 details in Everything

4. Dev Machine Guard v1.18.1 (2026-10-28) Oct 28, 2026 · issue 124

   Adds automatic MSI self-updates with rollback, configurable device identity via device_id in config.json, and a --disable-runtime-scan flag.

   • Opt out of automatic MSI self-updates at install time with AUTOUPDATE=0; when enabled, the agent verifies SHA-256SHA-256A cryptographic hash function that turns data into a fixed 256-bit fingerprint. It is used to check integrity; hashing does not encrypt the original data. hash, signed checksum, and Authenticode publisher before updating, rolls back automatically on failure, and never retries a bad version.
   • New supported platformTurn off language runtime scanning by setting enable_runtime_scan to false in the agent configuration or passing --disable-runtime-scan; scanning is on by default.
   • Set a custom device_id in the agent's config.json to give a device a stable, admin-controlled identity used by scans, scheduling, hooks, and device policies.

   3 details in Everything

   • New supported platform Windows system package inventory
   • New integration Package inventory for Go, Rust, PHP, .NET, and Ruby ecosystems
   • New integration GitHub Copilot plugin inventory in Agent Plugins
5. announced Oct 27, 2026 · issue 123

   StepSecurity adds org-wide AI agent plugin inventory for Claude Code and Codex — on developer machines and in GitHub repos.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢ ⤢
   └──▷ HOW TO FIND IT
   Identify which developer machines have a specific Claude Code plugin enabled and where its files actually came from — useful when investigating a suspicious marketplace listing.
   📍In Dev Machine Guard, go to IDE & AI Agents › Agent Plugins, search by plugin name or publisher, filter by source kind (e.g., github or local), and select the plugin to open its detail panel showing per-device installation scope, version, state, and marketplace vs. plugin file source.
   Determine which repositories in your org enable a published Claude Code plugin for all contributors — catching broad exposure before a compromised plugin update rolls out.
   📍In StepSecurity, go to GitHub › AI Agents › Agent Plugins, filter by agent (Claude Code), select the plugin, and review the Used by panel listing every repository whose .claude/settings.json enables the plugin.
   Export the full org-wide AI agent plugin inventory as CSV for offline review or SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents. ingestion.
   📍In Dev Machine Guard, go to IDE & AI Agents › Agent Plugins and click the CSV export button to download the full inventory including plugin names, publishers, source kinds, versions, and enabled/disabled device counts.
   • Adds a plugin detail panel in Dev Machine Guard listing marketplace source, plugin file origin, per-device installation scope (User, Project, Local, System), version, state, and component breakdown by kind (skills, commands, 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), with Claude Code skill usage counts.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Attributes plugin-sourced skills and MCP servers back to their originating plugin on the existing Agent Skills and MCP Servers pages, showing the marketplace and plugin that delivered each component.
   • Adds an Agent Plugins page under GitHub > AI Agents that inventories plugin manifests (.claude-plugin/plugin.json for Claude Code, .codex-plugin/plugin.json for Codex) published across your organization's GitHub repositories, identifying the publishing repository, plugin folder path, declared metadata, branch, commit scanned, and which repositories enable the plugin via their settings.
   • Exposes a Used by view in the GitHub Agent Plugins detail panel listing every repository whose settings enable a given plugin for all contributors, distinct from repositories that merely publish it.
   • Supports CSV export of the full Dev Machine Guard Agent Plugins inventory.

   +1 more show less

   • Supports filtering the Dev Machine Guard Agent Plugins inventory by installation scope (User, Project, Local, System), source kind, and enablement, and filtering the GitHub Agent Plugins inventory by agent or repository.
   • Adds a plugin detail panel in Dev Machine Guard listing marketplace source, plugin file origin, per-device installation scope (User, Project, Local, System), version, state, and component breakdown by kind (skills, commands, 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), with Claude Code skill usage counts.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Adds an Agent Plugins page under GitHub > AI Agents that inventories plugin manifests (.claude-plugin/plugin.json for Claude Code, .codex-plugin/plugin.json for Codex) published across your organization's GitHub repositories, identifying the publishing repository, plugin folder path, declared metadata, branch, commit scanned, and which repositories enable the plugin via their settings.
   • Exposes a Used by view in the GitHub Agent Plugins detail panel listing every repository whose settings enable a given plugin for all contributors, distinct from repositories that merely publish it.
   • Supports CSV export of the full Dev Machine Guard Agent Plugins inventory.
   • Supports filtering the Dev Machine Guard Agent Plugins inventory by installation scope (User, Project, Local, System), source kind, and enablement, and filtering the GitHub Agent Plugins inventory by agent or repository.

   1 more in Everything

6. Dev Machine Guard: Malicious Files Rules and Findings (2026-10-27) Oct 27, 2026 · issue 123

   StepSecurity Dev Machine Guard renames Suspicious Files to Malicious Files and adds dedicated Rules and Findings pages for fleet-wide detection management.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢
   └──▷ HOW TO FIND IT
   Suppress noisy detections fleet-wide by deactivating a rule from the Rules page, keeping findings visible under the Suppressed tab without deleting them.
   📍In Dev Machine Guard, go to Developer Machines › Rules, locate the rule by name or filter by campaign, then click Deactivate to stop it scanning every enrolled device and move its existing findings to the Suppressed tab. Click Activate to re-enable it when needed.
   Triage high-confidence malicious file hits across your fleet and export them for further analysis or ticketing.
   📍In Dev Machine Guard, go to Developer Machines › Findings, set the confidence filter to high, select the New tab to see unflagged items, then click Export to download a CSV of the filtered findings.
   • Adds a Rules page under Developer Machines listing every malicious-file detection rule with its campaign, indicator groups, and file patterns; supports search by rule name and filter by campaign.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Adds per-rule activate/deactivate controls that stop a rule from scanning all enrolled devices and move its findings to a Suppressed state, reversible at any time.
   • Adds a Findings page under Developer Machines showing flagged files grouped into New, Resolved, and Suppressed tabs, with device, user, OS, campaign, confidence, and status metadata.
   • Supports filtering Findings by confidence or campaign and exporting findings to CSV.

   4 details in Everything

7. Malicious Files for GitHub Repositories (Beta) (2026-10-27) Oct 27, 2026 · issue 123

   StepSecurity adds Malicious Files scanning (Beta) for GitHub repos — full history across branches, PRs, and tags.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢
   └──▷ HOW TO FIND IT
   Review active detection rules to understand which supply chain campaigns are covered and at what severity before interpreting findings.
   📍In the StepSecurity console, go to GitHub › Malicious Files › Rules to see each rule's campaign, severity, revision, and the ordered checks applied to every file version.
   Export all Findings for a repository to CSV for offline triage or ingestion into a SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents., filtered to high-severity results.
   📍In the StepSecurity console, go to GitHub › Malicious Files › Findings, filter by severity, then select Export to CSV.
   • Adds a Malicious Files section under GitHub scans that inspects full repository history — every branch, pull request, and tag — for files planted by known supply chain attack campaigns such as Shai-Hulud, GlassWorm, and Miasma.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Adds a Rules view showing each detection rule's campaign, severity, revision, and the ordered checks it runs against every file version.
   • Adds a Scan Status view showing the most recent scan per repository, including commits and file versions scanned, findings, and last/next scan times.
   • Adds a Findings view listing every file that matched a rule, with associated branches, pull requests, tags, matched checks, and the introducing commit — filterable by severity, campaign, or rule.
   • Enables export of Findings to CSV (Comma-Separated Values) format.

   5 details in Everything

8. Secure Registry: Inventory and Statistics (Beta) (2026-10-26) Oct 26, 2026 · issue 122

   StepSecurity Secure Registry gains two new beta pages — Inventory and Statistics — for package visibility and request analytics.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢
   └──▷ HOW TO FIND IT
   Identify every package version your teams have pulled that is affected by a specific advisory, so you can prioritise remediation.
   📍In Secure Registry, go to Inventory › search by advisory ID or alias › review the listed package versions, their fixed versions, and the upgrade guidance shown in each package detail view.
   Audit all packages requested under a specific open-source license by searching with its 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. identifier (e.g. GPL-3.0) in Inventory.
   📍In Secure Registry, go to Inventory › enter the SPDX license identifier (e.g. GPL-3.0) in the search box › apply License Status filter to review every requested version under that license.
   Review how many requests your controls blocked versus served over the past week to validate policy effectiveness.
   📍In Secure Registry, go to Statistics › select the desired time range › inspect the daily request history chart and the per-control breakdown of modified and blocked request counts.
   • Adds Secure Registry Inventory page listing every package requested through Secure Registry with its last pulled version, request counts, licenses, and known vulnerabilities.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Supports package search in Inventory by package name, SPDX license identifier, or advisory ID/alias to surface affected versions.
   • Adds filtering in Inventory by ecosystem, vulnerability severity, minimum CVSSCVSSCommon Vulnerability Scoring System, a method for describing a vulnerability's characteristics and severity. Its score is not the probability that attackers will exploit it. score, version, and license status.
   • Adds per-package detail view in Inventory showing first-seen date, last puller, advisories with fixed versions and upgrade guidance, and the last five allowed pulls.
   • Adds Secure Registry Statistics page showing requests handled over a selected time range, broken down into served-as-is, modified, and blocked, with a daily request history and per-control counts of modified and blocked requests.

   5 details in Everything

9. Agent Plugins Inventory for GitHub Repositories (2026-10-16) Oct 16, 2026 · issue 114

   StepSecurity adds an Agent Plugins inventory page that tracks Claude Code and Codex plugins published from GitHub repositories.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

   ⤢ ⤢ ⤢
   └──▷ HOW TO FIND IT
   Audit which Claude Code or Codex plugins your org's repos publish and what hooks they register — navigate to GitHub > AI Agents > Agent Plugins in the StepSecurity console.
   📍In the StepSecurity console, go to GitHub › AI Agents › Agent Plugins. Use the Filter by agent dropdown to select Claude Code or Codex, then search by component name to find plugins that register hooks on a specific event.
   Check which repos expose a specific plugin to all their contributors via the Used by view, to assess blast radius before modifying or removing a plugin.
   📍In the StepSecurity console, go to GitHub › AI Agents › Agent Plugins. Click a plugin name to open its details panel, then select the Used by tab to see every repository that has enabled it in its settings.
   • Adds an Agent Plugins page under GitHub > AI Agents that inventories Claude Code and Codex plugins detected from .claude-plugin/plugin.json or .codex-plugin/plugin.json manifests at the repository root or in subfolders.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
   • Displays per-plugin metadata — declared version, description, author, license, homepage, and repository — alongside the actual repository, branch, and commit StepSecurity scanned, with a direct link to each manifest.
   • Surfaces a Used by view showing which repositories in an organization enable a given plugin in their settings, making it available to all contributors in those repos.
   • Lists every skill and other component a plugin supplies, with hooks shown by the event they run on.
   • Supports filtering by agent or repository and searching by plugin or component name.

   5 details in Everything

10. Secure Registry: Packagist (PHP) Support (Beta) (2026-10-16) Oct 16, 2026 · issue 111

    Packagist Registry Settings panel adds Dist URL Rewriting, Keep Git Source, Block Branch Versions, and ::dev:/::gha: API-key source attribution for Composer.

    • Adds Dist URL Rewriting setting to route package zip downloads through Secure Registry so every archive is checked against your policy.
    • New integrationAdds Keep Git Source setting to preserve git clone entries in Composer package metadata.
    • Adds Block Branch Versions setting to hide branch versions from Composer package metadata.
    • Adds source attribution for Composer: appending ::dev: or ::gha: to the API key in http-basic or COMPOSER_AUTH traces requests to a developer machine or 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. run in Policy Evaluations.

    4 details in Everything

11. PAT Governance: Tenant-Wide Exemptions (2026-10-12) Oct 12, 2026 · issue 105

    StepSecurity PATPATA 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. Governance gains tenant-wide exemptions, letting you exempt a token or user across every org at once.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢ ⤢
    └──▷ HOW TO FIND IT
    Exempt a shared service account token from PAT Governance across all orgs in your tenant without touching each org individually.
    📍1. In the organization selector, choose All Organizations. 2. Navigate to Apps & PATs › Exemptions. 3. Click Add tenant-wide exemption and enter the token or user to exempt. 4. Confirm — the exemption now applies across every organization and appears labeled Tenant-wide in the unified exemptions list.
    • Adds tenant-wide exemptions in PAT Governance: select All Organizations in the org selector, open Apps & PATsPATsA 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. > Exemptions, and click Add tenant-wide exemption to exempt a token or user across every organization in the tenant simultaneously.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds a unified exemptions view under All Organizations that lists every organization's exemptions in one place, with tenant-wide exemptions labeled Tenant-wide.

    2 details in Everything

12. Secure Registry: RubyGems and Cargo Support, and Custom Block Messages for npm (2026-10-12) Oct 12, 2026 · issue 105

    Secure Registry adds RubyGems and Cargo support, plus custom block messages and hidden-version notices for npm.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢ ⤢
    └──▷ HOW TO FIND IT
    Tell developers how to request a package exception when a control blocks their npm install, using the package name in the message.
    📍In npm Registry Settings, set Block Message to: 'Package {{package}} was blocked by {{control}}. To request an exception, open a ticket at https://security.example.com/exceptions.'
    • Adds RubyGems support: point Bundler, the gem CLI, or an artifact repository manager at https://registry.stepsecurity.io/ruby to evaluate every gem request against your RubyGems policy, with Cooldown Period, Compromised Packages, and Custom Block List controls; integration paths cover JFrog Artifactory, Sonatype Nexus Repository, and Direct.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds Cargo support: point Cargo or JFrog Artifactory at the sparse index sparse+https://registry.stepsecurity.io/rust/ to evaluate every crate request against your Cargo policy, with Cooldown Period, Compromised Packages, and Custom Block List controls; integration paths cover JFrog Artifactory and Direct, and existing Cargo.lock files keep working without being regenerated.
    • Adds a Block Message setting in npm Registry Settings to include custom text (up to 500 characters) in the error developers see when a control blocks a request, supporting placeholders such as {{control}} and {{package}}; applies to every control except Cooldown Period.
    • Adds a Hidden Versions Notice setting for npm that replaces the default notice shown when package versions are hidden by policy, accepting up to 300 characters and supporting placeholders such as {{count}} and {{package}}.
    • Adds RubyGems support: point Bundler, the gem CLI, or an artifact repository manager at https://registry.stepsecurity.io/ruby to evaluate every gem request against your RubyGems policy, with Cooldown Period, Compromised Packages, and Custom Block List controls; integration paths cover JFrog Artifactory, Sonatype Nexus Repository, and Direct.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds Cargo support: point Cargo or JFrog Artifactory at the sparse index sparse+https://registry.stepsecurity.io/rust/ to evaluate every crate request against your Cargo policy, with Cooldown Period, Compromised Packages, and Custom Block List controls; integration paths cover JFrog Artifactory and Direct, and existing Cargo.lock files keep working without being regenerated.

    2 more in Everything

13. Harden Runner Releases: October 9, 2026 (2026-10-09) Oct 9, 2026 · issue 102

    Harden-Runner gains GitHub Enterprise Server support, macOS deny-list egress, and namespace-scoped 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. for ARC on Kubernetes.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    └──▷ USE IT
    Run Harden-Runner on a GitHub Enterprise Server self-hosted ephemeral VM when baking the agent into the image is not possible.
    .github/workflows/your-workflow.yml copy

    - uses: step-security/harden-runner@v2.22.1
      with:
        deploy-on-self-hosted-vm: true
        customer: your-stepsecurity-customer
        server-name: your-ghes-server
        egress-policy: audit

    Block specific egress destinations on a macOS runner while leaving all other outbound traffic reachable.
    .github/workflows/your-workflow.yml copy

    - uses: step-security/harden-runner@v2.22.1
      with:
        egress-policy: block
        denied-endpoints: 'example.com:443 suspicious-host.io:80'

    Install ARC Harden Runner on a multi-tenant Kubernetes cluster without cluster-scoped RBAC resources.
    yaml copy

    rbac:
      scope: namespaced

    • Adds customer and server-name inputs to the step-security/harden-runner Action (v2.22.0) to enable Harden-Runner on GitHub Enterprise Server environments, required for TLS inspection, policy fetching, and agent deployment.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds deploy-on-self-hosted-vm: true input to deploy the agent directly on an ephemeral self-hosted Linux VM at workflow runtime, for cases where baking the agent into the VM image is not possible.
    • Extends the denied-endpoints input to macOS runners, enabling deny-list egress policy on macOS to block specific destinations while leaving all others reachable.
    • Adds rbac.scope: namespaced to ARC Harden Runner (v2.21.0) to install with only Role and RoleBinding objects in the release and runner namespaces, with no ClusterRole or ClusterRoleBinding, for multi-tenant Kubernetes clusters.
    • Brings Harden-Runner agent (Self-Hosted VM v1.9.3) to self-hosted runners connected to GitHub Enterprise Server when installed via the Harden-Runner GitHub Action.
    • Adds customer and server-name inputs to the step-security/harden-runner Action (v2.22.0) to enable Harden-Runner on GitHub Enterprise Server environments, required for TLS inspection, policy fetching, and agent deployment.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds deploy-on-self-hosted-vm: true input to deploy the agent directly on an ephemeral self-hosted Linux VM at workflow runtime, for cases where baking the agent into the VM image is not possible.
    • Adds rbac.scope: namespaced to ARC Harden Runner (v2.21.0) to install with only Role and RoleBinding objects in the release and runner namespaces, with no ClusterRole or ClusterRoleBinding, for multi-tenant Kubernetes clusters.
    • Brings Harden-Runner agent (Self-Hosted VM v1.9.3) to self-hosted runners connected to GitHub Enterprise Server when installed via the Harden-Runner GitHub Action.

    1 more in Everything

14. announced Oct 8, 2026 · issue 103

    StepSecurity adds AI agent skill inventory for developer machines and GitHub repos, with executable-content flags and CSV export.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢ ⤢ ⤢
    └──▷ HOW TO FIND IT
    Immediately surface every repository skill that pre-approves shell execution — the highest-risk class, equivalent in impact to a CI/CD workflow change.
    📍In the StepSecurity dashboard, go to GitHub › Agent Skills, then apply the Shell execution filter. Open each flagged skill's details panel and review Declared behavior › Allowed tools for entries such as Bash.
    Find developer machines where the same skill has diverging content — a signal that one copy may have been tampered with or is outdated.
    📍In the StepSecurity dashboard, go to IDE & AI Agents › Agent Skills (Dev Machine Guard). Sort or filter by Content hashes › 1, then select the skill to open the per-device panel grouped by content hash and compare variants.
    Audit unmanaged (Local) skills on developer machines that have no upstream source in skills.sh or a plugin marketplace — the class most likely to have been hand-planted.
    📍In the StepSecurity dashboard, go to IDE & AI Agents › Agent Skills (Dev Machine Guard) and filter by Source = Local. Export the result as CSV for offline review.
    • Adds an Agent Skills page in Dev Machine Guard (under IDE & AI Agents) that inventories AI agent skills across developer machines, scanning ~/.agents/skills, ~/.claude/skills, project-level skill folders, and skills.sh lock files across 18 AI coding agents including Claude Code, Codex, GitHub Copilot, Cursor, and Gemini CLI; adds code, hooks, and shell flags per skill to surface executable content, plus a command flag for Claude Code commands, filterable via Has code, Has hooks, and Shell execution filters in both inventory views; adds ProvenanceProvenanceEvidence recording where a software artifact came from and how it was built, including its source and build process. It helps a consumer check that the artifact matches a trusted build. tracing for skills.sh-managed skills, surfacing upstream source, ref, path, upstream folder hash, and the skills-lock.json file; supports CSV export of the full Dev Machine Guard agent skills inventory; adds notifications for new agent skills detected after a device's initial baseline, with per-device alerts; and surfaces content-hash comparison across devices so practitioners can identify installations where the same skill's SKILL.md content differs between machines.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds an Agent Skills page under GitHub that scans the default branch of each non-archived repository in your organization daily, detecting skills in folders such as .claude/skills, .cursor/skills, .windsurf/skills, and .agents/skills; adds a Declared Behavior panel for GitHub repository skills showing frontmatter details including pre-approved tools (e.g., Read, Write, Bash), invocation restrictions (no model invocation, not user invocable), forked subagent context, and model overrides; and labels repository skills with invalid yaml when frontmatter cannot be parsed, and symlink when the skill folder is reached through a symbolic link.

    2 details in Everything

15. Agent Skills Inventory for GitHub Repositories (2026-10-05) Oct 5, 2026 · issue 096

    StepSecurity adds a repository-level Agent Skills inventory page showing AI agent skills, executable flags, declared behavior, and provenanceprovenanceEvidence recording where a software artifact came from and how it was built, including its source and build process. It helps a consumer check that the artifact matches a trusted build..⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Review which repository skills ship executable scripts or inject shell commands before allowing agent sessions to load them.
    📍In the StepSecurity console, go to GitHub Inventories › Agent Skills, then use the Flag filter to select shell or code to surface only skills with executable content.
    Audit provenance of a skills.sh-managed skill to confirm its upstream source and lock file before trusting it in CI.
    📍In the StepSecurity console, go to GitHub Inventories › Agent Skills, open the details panel for a skill, and check the upstream source, ref, upstream folder hash, and the skills-lock.json record shown under Provenance.
    • Adds an Agent Skills page under GitHub inventories listing every AI agent skill committed to the default branch of each org repository, refreshed daily, with per-skill detail on repository, loading agent (Claude Code, Cursor, Windsurf, or shared .agents/skills directory), path, and file and script counts.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Supports filtering the Agent Skills inventory by source, agent, repository, or flag, and searching by skill name.
    • Flags skills with executable content markers: code (executable scripts), hooks (declared hooks block), shell (injected shell commands on skill load), invalid yaml (unparseable SKILL.md frontmatter), and symlink (skill reached through a symbolic link).
    • Adds a details panel showing each skill's pre-approved tools (allowed-tools), who can invoke it, whether it runs in a forked context, and any model override it sets.
    • Provides provenance for skills managed by the skills.sh CLI, tracing each skill to its upstream source, ref, upstream path, upstream folder hash, and the skills-lock.json installation record.

    5 details in Everything

16. Harden Runner Releases: September 30, 2026 (2026-09-30) Sep 30, 2026 · issue 090

    Egress policy enforcement expands to self-hosted Windows (deny list) and macOS (allow list + deny list, block mode) runners.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    • Adds deny list egress policy to Self-Hosted Windows Harden Runner v0.10.0, blocking specific destinations while leaving all others reachable.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    1 detail in Everything

17. PAT Governance: Exemptions, Configurable PAT Abuse Thresholds, GitHub Audit Log, and Classic PATs on GitHub Enterprise Server (2026-09-29) Sep 29, 2026 · issue 090

    StepSecurity adds PATPATA 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. exemptions, configurable abuse thresholds, GitHub audit log review, and classic PAT inventory on GHES.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Grant a temporary exemption for every token a specific user holds — useful when a developer is mid-incident and automated revocation would disrupt response.
    📍In the StepSecurity console, go to Apps & PATs › Exemptions, click Add Exemption, select the user, choose All tokens (current and future), and set the expiry date.
    Tune PAT Abuse detection so noisy rules don't fire until enough baseline data is collected, reducing false-positive alerts for a newly on-boarded organization.
    📍In the StepSecurity console, go to Apps & PATs › PAT Abuse › Configure, toggle individual rules on or off, and set the learning period and threshold values for each rule.
    Investigate which workflow runs were deleted and by whom after a suspected tampering event, using the 10-day retained audit log.
    📍In the StepSecurity console, go to GitHub › Audit Log › Deleted runs and filter by date range and actor to identify who deleted runs.
    • Adds an Exemptions tab on Apps & PATs to exempt a single token or all tokens a user holds (now and future) for a fixed period, skipping policy checks, pre-expiry reminders, and automated revocation until the exemption expires — with automatic re-enrollment afterward.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Surfaces exemptions created via ServiceNow or through the API on the same Exemptions tab.
    • Adds a Configure panel on the PAT Abuse tab to turn each abuse detection rule on or off and set its learning period and thresholds; changes apply from the next audit-log poll and preserve already-learned baselines.
    • Adds a new Audit Log page under GitHub showing organization GitHub audit log activity refreshed every 15 minutes and retained for 10 days, with a PAT activity tab summarizing calls per token and a Deleted runs tab showing deleted workflow runs.
    • Supports classic PAT inventory on GitHub Enterprise Server by connecting a site_admin-scoped token from a dedicated site administrator account to enumerate classic PATsPATsA 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. for organization members and outside collaborators.
    • Adds an Exemptions tab on Apps & PATs to exempt a single token or all tokens a user holds (now and future) for a fixed period, skipping policy checks, pre-expiry reminders, and automated revocation until the exemption expires — with automatic re-enrollment afterward.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds a new Audit Log page under GitHub showing organization GitHub audit log activity refreshed every 15 minutes and retained for 10 days, with a PAT activity tab summarizing calls per token and a Deleted runs tab showing deleted workflow runs.
    • Supports classic PAT inventory on GitHub Enterprise Server by connecting a site_admin-scoped token from a dedicated site administrator account to enumerate classic PATsPATsA 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. for organization members and outside collaborators.

    2 more in Everything

18. Dev Machine Guard v1.17.0 (2026-09-29) Sep 29, 2026 · issue 089

    Dev Machine Guard v1.17.0 adds PyPI, uv, and Go Secure Registry policies plus a Claude Code/Codex Agent Plugins inventory and Insomnia credential scanning.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Audit which Claude Code and Codex plugins are installed across your fleet before approving them for use in a compliance review.
    📍In the StepSecurity console, go to IDE & AI Agents › Agent Plugins to view each plugin's marketplace source, publisher, enabled state per device, and the MCP servers it supplies.
    Roll out a scoped private registry for your org's npm packages alongside the Secure Registry, using an environment-variable token that StepSecurity never stores.
    .npmrc copy

    @myorg:registry=https://npm.mycompany.internal
    //npm.mycompany.internal/:_authToken=${MYORG_NPM_TOKEN}

    • Adds PyPI package config policies that point pip and uv at your tenant's Secure Registry by setting index in pip.conf and uv.toml, storing a shared credential in ~/.netrc, and restoring prior config when the policy is detached.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds Go package config policies that set the Secure Registry as the sole GOPROXYGOPROXYAn environment variable and protocol used by the Go toolchain to specify a module proxy server from which `go` commands fetch dependencies, enabling reproducible builds, caching, and access control over third-party packages. in the user's go/env file via the same shared ~/.netrc credential, leaving GOPRIVATE, GONOPROXY, GONOSUMDB, and GOSUMDB untouched.
    • Adds support for an optional list of extra .npmrc settings in npm package config policies, delivered to every assigned device inside the managed block, with tokens referenced as ${ENV_NAME} from the developer's environment.
    • Adds a new Agent Plugins inventory page under IDE & AI Agents that tracks Claude Code and Codex plugins fleet-wide, including marketplace/repository source, publisher, installed and enabled state per device, and the skills, commands, 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 each plugin supplies.
    • Adds Insomnia to the credential inventory under a new API clients category, covering request authentication, credential headers, environment and secret variables, OAuth tokens, request history, and account/provider stores — secret variables in Insomnia's encrypted storage are reported as Protected.

    +3 more show less

    • Adds detection for Grok Build, Kimi Code, Muse Code, Hermes Agent, Oh My Pi as Coding Tools and Kiro as an IDE, with their skills plus Kiro, Windsurf, Antigravity, and OpenClaw skills appearing under Agent Skills.
    • Agent Skills now lists standalone Claude Code commands and shows a per-skill usage count drawn from Claude Code's recorded history.
    • Redesigns the profile Compliance tab to show a compliance card per policy, one overall status per device, a device panel with per-policy status and reason, and highlights devices whose agent is too old for a policy.
    • Adds PyPI package config policies that point pip and uv at your tenant's Secure Registry by setting index in pip.conf and uv.toml, storing a shared credential in ~/.netrc, and restoring prior config when the policy is detached.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds Go package config policies that set the Secure Registry as the sole GOPROXYGOPROXYAn environment variable and protocol used by the Go toolchain to specify a module proxy server from which `go` commands fetch dependencies, enabling reproducible builds, caching, and access control over third-party packages. in the user's go/env file via the same shared ~/.netrc credential, leaving GOPRIVATE, GONOPROXY, GONOSUMDB, and GOSUMDB untouched.
    • Adds a new Agent Plugins inventory page under IDE & AI Agents that tracks Claude Code and Codex plugins fleet-wide, including marketplace/repository source, publisher, installed and enabled state per device, and the skills, commands, 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 each plugin supplies.
    • Adds Insomnia to the credential inventory under a new API clients category, covering request authentication, credential headers, environment and secret variables, OAuth tokens, request history, and account/provider stores — secret variables in Insomnia's encrypted storage are reported as Protected.

    4 more in Everything

19. Low Score Exemptions for Control Evaluation (2026-09-28) Sep 28, 2026 · issue 088

    StepSecurity lets you exempt specific 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. from the low-score control by name or owner wildcard.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Suppress low-score alerts for a vetted Action your team has reviewed, without needing to pin a specific version.
    📍In the StepSecurity console, go to Settings › Control Evaluation, open the Control options card for 'Third-party GitHub Actions with high scores should be used', and add example-org/example-action (or example-org/* to cover all Actions under that owner) to the exemptions list.
    • Adds exemption support for the 'Third-party GitHub Actions with high scores should be used' control, configurable under Settings > Control Evaluation in the Control options card, allowing Actions to be exempted by exact name (e.g.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ example-org/example-action) or owner wildcard (e.g. example-org/*).
    • Exemptions are version-independent — any version ref in an exemption entry is ignored, so the exemption covers every version of the Action.
    • Adds exemption support for the 'Third-party GitHub Actions with high scores should be used' control, configurable under Settings > Control Evaluation in the Control options card, allowing Actions to be exempted by exact name (e.g.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ example-org/example-action) or owner wildcard (e.g. example-org/*).

    1 more in Everything

20. ARC Harden Runner v2.20.0 (2026-09-23) Sep 23, 2026 · issue 083

    ARC Harden Runner v2.20.0 adds keyless AWS Secrets Manager API key fetch via IRSAIRSAAn AWS mechanism that lets a Kubernetes pod assume an IAM role via a service account, so tools running in EKS can call AWS APIs without embedding long-lived credentials., custom agent env vars, and configurable TetragonTetragonAn open-source runtime security and observability tool from Isovalent that uses eBPF to enforce policies and trace system calls, network activity, and process behavior directly in the Linux kernel. security context.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    └──▷ USE IT
    Eliminate the static API key from Helm values by having the daemonset pull it from AWS Secrets Manager via the pod's 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. role — useful in environments where secrets in Helm values are a compliance risk.
    values.yaml copy

    serviceAccount:
      annotations:
        eks.amazonaws.com/role-arn: arn:aws:iam::<account-id>:role/<irsa-role>
    env:
      customer: "<your-org>"
      keyless:
        enabled: "true"
        region: "us-east-1"
        secretName: "stepsecurity/orgs/<github-org>/arc-api-key"
        refreshIntervalHours: "3"

    Route all Harden Runner agent traffic through a corporate HTTP proxy in air-gapped or proxied 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. clusters.
    values.yaml copy

    env:
      customEnvVars:
        - name: HTTP_PROXY
          value: "http://proxy.corp.example.com:3128"
        - name: HTTPS_PROXY
          value: "http://proxy.corp.example.com:3128"
        - name: NO_PROXY
          value: "169.254.169.254,10.0.0.0/8"

    • Adds keyless.enabled, keyless.region, keyless.secretName, and keyless.refreshIntervalHours Helm values to let the Harden Runner daemonset fetch its StepSecurity API key from AWS Secrets Manager via IRSA (pod IAM role) instead of a static Helm value, refreshing on a configurable schedule (default 3 hours); the secret must be a JSON object with an api_key field and the IAM role requires secretsmanager:GetSecretValue.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠
    • Adds env.customEnvVars Helm value to inject arbitrary environment variables (e.g. HTTP_PROXY, HTTPS_PROXY, NO_PROXY) into the harden-runner container for proxied cluster deployments.
    • Adds tetragon.containerSecurityContext Helm value to configure the Tetragon sidecar's security context.

    3 details in Everything

21. Apps & PATs Inventory for GitHub Enterprise Server (2026-09-18) Sep 18, 2026 · issue 082

    StepSecurity Apps & PATsPATsA 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. Inventory now available for GitHub Enterprise Server organizations.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
22. PAT Governance: Reminder Recipients and Custom Email Content (2026-09-18) Sep 18, 2026 · issue 082

    StepSecurity PATPATA 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. Governance now supports CC recipients and per-class, per-band custom email content for pre-expiry token reminders.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Route all PAT expiry reminders to a security team inbox in addition to the org notification address, and set a tighter fine-grained-only schedule.
    📍In the StepSecurity console, go to Settings › Control Evaluation › Reminders & Automation (PAT Governance Controls). Under CC Recipients, add your security distribution list address. Under Reminder Schedule, enable the fine-grained override and set custom days (e.g. 14, 3, 1) to replace the shared 30/7/1-day default for fine-grained tokens only.
    Customize the 1-day pre-expiry email body for classic tokens to include your internal rotation runbook URL.
    📍In the StepSecurity console, go to Settings › Control Evaluation › Reminders & Automation (PAT Governance Controls). Select the classic token class and the 1 day reminder band, then edit the subject and body fields — use placeholders such as {{owner}}, {{token_name}}, {{organization}}, {{days_left}}, {{expiry_date}}, and {{management_url}} — and save. Fields left unchanged keep the built-in email.
    • Adds CC recipient support so additional addresses (e.g.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ a security distribution list) are copied on every pre-expiry PAT reminder sent by the organization.
    • Adds per-token-class reminder schedules with optional classic and fine-grained overrides that replace the shared default schedule (30, 7, 1 days before expiry) for one class.
    • Adds per-token-class and per-reminder-band subject and body editors, prefilled with the default email, supporting placeholders for owner, token name, organization, days left, expiry date, and the management URL.
    • Adds CC recipient support so additional addresses (e.g.⁠︀︄︇︀︅︉︀︅︀︀︃︀︀︅︅︀︅︊︀︄︆︀︃︄⁠ a security distribution list) are copied on every pre-expiry PAT reminder sent by the organization.
    • Adds per-token-class and per-reminder-band subject and body editors, prefilled with the default email, supporting placeholders for owner, token name, organization, days left, expiry date, and the management URL.

    1 more in Everything

23. Webhook Integration: Multiple Endpoints per Tenant (2026-09-07) Sep 7, 2026 · issue 068

    StepSecurity now supports up to ten webhooks per tenant, each with independent endpoints, event subscriptions, and repository scopes.

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Route runtime detections to your SIEMSIEMSecurity information and event management, collecting logs from many systems and correlating events to find suspicious activity. Analysts use it to investigate alerts and incidents. and audit logs to a separate log archive without either receiver filtering unwanted traffic.
    📍In the StepSecurity console, go to Webhook Integration and click Add Webhook. Configure the first webhook with endpoint URL pointing to your SIEM, subscribe it to Detections and Insights, and set the Envelope template and auth headers your SIEM expects. Click Add Webhook again, set the second endpoint URL to your log archive, subscribe it to Secure Registry Audit Logs and API Audit Logs, then click Save changes.
    • Supports up to ten webhooks per tenant, each configured with its own endpoint URL, HTTP method, headers, message format, and Envelope template — replacing the previous single-webhook-per-tenant limit.
    • Adds per-webhook event subscriptions choosing from four event types: Insights, Detections, Secure Registry Audit Logs, and API Audit Logs, in any combination.
    • Adds per-webhook repository scope, so one endpoint can be scoped to a subset of repositories while another covers all current and future repositories.
    • Delivers independent copies of each event to every subscribed webhook — a failure at one endpoint does not affect delivery to others.
    • Adds an Add Webhook control in the UI to create additional webhook endpoints, each rendered as a collapsible card showing its name, endpoint, and event-type chips.
    • Adds per-webhook event subscriptions choosing from four event types: Insights, Detections, Secure Registry Audit Logs, and API Audit Logs, in any combination.
    • Adds per-webhook repository scope, so one endpoint can be scoped to a subset of repositories while another covers all current and future repositories.
    • Delivers independent copies of each event to every subscribed webhook — a failure at one endpoint does not affect delivery to others.

    2 more in Everything

24. announced Sep 2, 2026 · issue 064

    Harden-Runner v2.21.0 adds denied-endpoints input for deny-list egress policies, blocking named destinations while leaving all others reachable.

    ⤢ ⤢
    └──▷ USE IT
    Block direct npm and PyPI registry access org-wide to enforce routing through a package proxy, while leaving all other outbound traffic unrestricted.
    .github/workflows/ci.yml copy

    - uses: step-security/harden-runner@v2.21.0
      with:
        egress-policy: block
        denied-endpoints: |
          registry.npmjs.org
          pypi.org
          files.pythonhosted.org

    • Adds denied-endpoints input to the step-security/harden-runner GitHub Action (v2.21.0) — list domains, wildcard domains, or IP addresses to block, while leaving all unlisted destinations reachable under egress-policy: block.
    • Adds denied_endpoints input to the Self-Hosted VM Harden Runner (v1.9.0) for the same deny-list capability on self-hosted VMs.
    • Supports deny-list policies in the Policy Store attached at organization or cluster scope (ARC Harden Runner v2.19.0+), enforcing CI egress restrictions across all repositories without per-workflow-file edits.
    • Wildcard domain entries (e.g. *.example.org) in denied-endpoints block all subdomains with a single line, consistent with allow-list wildcard syntax.

    4 details in Everything

25. announced Sep 2, 2026 · issue 064

    Harden-Runner now secures 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. jobs on AWS CodeBuild-hosted runners (EC2EC2Amazon's cloud virtual-machine service, letting tools run on-demand servers without managing physical hardware. compute, managed and custom images) from v2.20.1.

    ⤢
    └──▷ USE IT
    Secure a CodeBuild-hosted GitHub Actions job to monitor all outbound network calls and process executions from inside your AWS 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..
    .github/workflows/build.yml copy

    jobs:
      build:
        runs-on: codebuild-my-project-${{ github.run_id }}-${{ github.run_attempt }}
        steps:
          - uses: step-security/harden-runner@v2.21.0
            with:
              egress-policy: audit
          - uses: actions/checkout@v4

    • Adds support for GitHub Actions runners hosted on AWS CodeBuild on EC2 compute, starting with step-security/harden-runner@v2.20.1, enabling outbound network, file-write, and process-execution monitoring for CodeBuild-hosted jobs.
    • Extends CodeBuild support in step-security/harden-runner@v2.21.0 to Linux containers on EC2 compute, covering custom images from Amazon ECRECRAmazon Elastic Container Registry, AWS's managed service for storing and distributing Docker and OCI container images, giving cyber tools a private, IAM-gated registry without self-hosting infrastructure. and external Docker registries.
    • Requires the CodeBuild project to be configured with Compute: EC2, Host kernel: kernel-6 (Amazon Linux 2023), and Privileged: Enabled — set in the project's Additional configuration panel — for the Harden-Runner agent to function.
    • Surfaces memory-read detections (the technique behind CVE-2025-8217) as a runtime alert with a triage runbook attached, for CodeBuild-hosted runner jobs.
    • Unifies CodeBuild-hosted, GitHub-hosted, ARC, and self-hosted runner fleets under a single dashboard, Policy Store, and detection baseline.

    5 details in Everything

26. PAT Governance for GitHub and Azure DevOps (2026-09-02) Sep 2, 2026 · issue 064

    StepSecurity adds continuous PATPATA 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. governance for GitHub and Azure DevOps with hourly policy evaluation, scope risk classification, and a new Identities & PATsPATsA 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. inventory page.

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Review the new Identities & PATs inventory for your Azure DevOps organization to spot over-privileged or stale tokens, service connections, and Entra service principals at a glance.
    📍In the StepSecurity console, go to Identities & PATs (Azure DevOps), select the Personal Access Tokens tab, click Refresh to pull the latest data from the Azure DevOps APIs, then click Export to download the token inventory for that tab.
    Triage PATs currently breaching your token policy — including over-privileged classic GitHub tokens with repo or admin:org scope — and revoke offenders directly from the dashboard.
    📍In the StepSecurity console, go to All Controls dashboard, filter by Token Controls, review each finding's evidence and impact, then click Revoke on any token that breaches the policy.
    • New Identities & PATs page for Azure DevOps inventories 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., service connections, and Entra service principals in one place, with hourly auto-refresh, on-demand refresh, and per-tab export.
    • Token controls evaluate your entire PAT inventory hourly on both GitHub and Azure DevOps organizations, surfacing breaching tokens as findings on the All Controls dashboard with evidence, impact, and a per-token revoke action.
    • Configurable max-age and idle-window thresholds per organization, with separate limits for fine-grained and classic GitHub tokens, and a stricter limit for high-privilege Azure DevOps tokens.
    • Pre-expiry reminder notifications sent to token owners at configurable bands — 30, 7, and 1 days by default — for both GitHub and Azure DevOps.
    • Scope risk classification on Azure DevOps flags tokens against the 35 Microsoft-rated high-privilege scopes, with colour-coded scope chips; on GitHub, coarse classic scopes (repo, admin:org, workflow) and fine-grained tokens combining all-repository access with write permissions are flagged.

    +1 more show less

    • Per-control notification events fire once when a token first breaches policy rather than on every scan, and each control's notifications are selectable independently.

    6 details in Everything

27. announced Aug 27, 2026 · issue 058

    Dev Machine Guard gains a Credentials inventory page showing plaintext credential exposure across developer fleets, available from agent v1.16.0.

    ⤢ ⤢ ⤢
    └──▷ HOW TO FIND IT
    When a credential-stealing package is disclosed, confirm which sources existed on affected machines so rotation is targeted rather than fleet-wide.
    📍In the StepSecurity dashboard, go to Dev Machine Guard › Credentials, filter by the relevant category (e.g. Cloud or Package registries), identify affected devices and entry counts, then export the view as CSV to drive a targeted rotation plan.
    • Adds a Credentials page in the StepSecurity dashboard that aggregates credential source coverage fleet-wide, showing per-source device count, plaintext vs. protected breakdown, location count, and individual entry count — filterable by category (Cloud, Source control, Package registries, Containers, Infrastructure) and exportable as CSV.
    • Covers thirteen credential source types across five categories, including ~/.npmrc, ~/.pypirc, ~/.vault-token, ~/.netrc, ~/.aws/credentials, ~/.aws/config, ~/.docker/config.json, gcloud application default credentials, kubeconfig, ~/.ssh private key files (id_rsa, id_ed25519, id_ecdsa), and ~/.git-credentials.
    • Per-source detail view lists every affected device labelled Plaintext, Protected, or both, with per-device summaries (e.g. '3 locations · 3 entries (2 plaintext)') and direct links to each machine.
    • On macOS, Full Disk Access must be granted to the agent via MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies. to reach TCC-protected credential paths and return complete inventory results.
    • Available from agent v1.16.0; existing Dev Machine Guard deployments gain the Credentials page automatically after each enrolled device completes its next scan.

    5 details in Everything

28. announced Aug 27, 2026 · issue 058

    Dev Machine Guard v1.16.0 adds a Browser Extensions inventory page covering Chrome, Edge, and Firefox across your developer fleet.

    ⤢ ⤢ ⤢ ⤢
    └──▷ HOW TO FIND IT
    During an incident, search by extension ID in the Browser Extensions page to instantly see which devices are running a compromised extension and at which versions.
    📍In the StepSecurity dashboard, go to Dev Machine Guard › Browser Extensions › search the extension ID in the table › review the per-device version grouping for affected machines.
    • Adds a Browser Extensions page in the StepSecurity dashboard showing a fleet-wide table of every extension detected, including name, extension ID, browser, store, version, device count, and last-seen timestamp — filterable by browser, capability, store, and engine, with an Export CSV download.
    • Rolls raw browser permissions into ten named capabilities — Global website access, Global request observation, Script injection, Global script injection, Cookie access, Global cookie access, Browsing history, Debugger access, Clipboard read, and Extension management — reflecting what the browser is currently honouring, not just what the extension declared in its manifest.
    • Per-device detail panel lists affected devices grouped by version (first 500 devices shown), enabling rapid scoping when an advisory names a specific bad version.
    • Covers Chrome, Edge, and Firefox on macOS, Windows, and Linux, including Firefox under snap and Flatpak and Edge under Flatpak; available from agent v1.16.0 with no changes required for existing Dev Machine Guard deployments.

    4 details in Everything

29. Dev Machine Guard Notifications (2026-08-26) Aug 26, 2026 · release notes

    Dev Machine Guard now supports tenant notifications for seven new detection events with per-device granularity.

    └──▷ HOW TO FIND IT
    Configure tenant-level alerts for specific detection types like suspicious file or 'package manager configuration change'.
    📍In the StepSecurity console, navigate to Notifications › Dev Machine Guard and enable individual toggles for new events.
    • Adds Seven new notification events to Dev Machine Guard: A new IDE, a new IDE extension, a new AI agent, a new 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, a new agent skill, a suspicious file, and a package manager configuration change.

    1 detail in Everything

30. ARC Harden Runner v2.19.0 (2026-08-24) Aug 24, 2026 · issue 055

    ARC Harden Runner v2.19.0 adds deny list egress enforcement and configurable image pull policy via Helm values.

    ⤢ ⤢
    └──▷ USE IT
    Stop ARC Harden Runner from pulling images on every pod start in air-gapped or registry-constrained clusters by setting pull policy to IfNotPresent in your Helm values file.
    values.yaml copy

    env:
      imagePullPolicy: IfNotPresent
    tetragon:
      imagePullPolicy: IfNotPresent

    • Adds env.imagePullPolicy and tetragon.imagePullPolicy Helm values to control image pull behavior for ARC Harden Runner and Tetragon images, enabling IfNotPresent for clusters that pre-pull images or operate behind constrained registries.
    • Extends deny list egress policy enforcement to ARC Harden Runner — policies attached at Cluster scope in the Policy Store now block specific destinations across every job in that cluster without requiring workflow file changes.

    2 details in Everything

31. Dev Machine Guard v1.16.0 (2026-08-21) Aug 21, 2026 · issue 052

    Dev Machine Guard v1.16.0 adds credential-location and browser-extension inventories, plus expanded AI agent 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. detection.

    ⤢ ⤢
    • New credential location inventory scans thirteen sources — AWS credentials, AWS config, GCP application default credentials, SSH private keys, git credential store, .netrc, GitHub CLI hosts, .npmrc, .pypirc, Docker config, kubeconfig, Terraform credentials, and Vault token — reporting exact file path, credential count, usable-or-wrapped status, file mode (macOS/Linux), and whether the file is tracked inside a git repository; no credential value, substring, digest, or fingerprint ever leaves the device.
    • New browser extension inventory covers Google Chrome, Microsoft Edge, and Mozilla Firefox (including Firefox under snap/flatpak and Edge under flatpak) on macOS, Windows, and Linux, recording per-extension identity (id, name, version, manifest version), enabled state and who disabled it, install method (store, sideloaded, registry installed, unpacked, or policy installed), store attribution and listing/flag status, Firefox signature state, and the host and API permissions the browser is currently honouring.
    • Pi, Factory Droid, and Amp are now detected in the AI agent and tool inventory, proven from on-disk artifacts (npm or Bun manifests, installer anchor directories, Homebrew cask roots, WinGet publisher directories, or pacman file manifests) rather than by binary name, with version read from disk or reported as unknown.
    • Detection of global installs now searches per-version trees of nvm, fnm, mise, volta, and asdf.
    • Two new agent skills roots are scanned: ~/.config/amp/skills and ~/.agent/skills.

    +3 more show less

    • Copilot CLI installed via gh copilot is now detected across seven additional install locations (the gh copilot download directories on both platforms, the non-root install script path, WinGet and npm global shims, and the gh-copilot extension) when a PATH lookup misses.
    • OpenCode MCP servers are now collected from ~/.config/opencode/opencode.json and opencode.jsonc (both formats, including comments and trailing commas), using the top-level mcp key OpenCode declares rather than mcpServers; project-level configs are found by the existing bounded search.
    • Secrets held in MCP server command, argument, and URL fields are now redacted before the MCP inventory is reported.

    8 details in Everything

32. Mint StepSecurity API Tokens from AWS Workloads (2026-08-20) Aug 20, 2026 · issue 055

    StepSecurity 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. federation now supports AWS, letting 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., EC2EC2Amazon's cloud virtual-machine service, letting tools run on-demand servers without managing physical hardware., and Lambda workloads mint API tokens without stored keys.

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Configure an org-scope trust policy in StepSecurity that allows any 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. role in a specific AWS account to mint API tokens — useful for giving an EKS cluster-wide 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. access without storing a key.
    📍In the StepSecurity console, go to Organization › Trust Policies › New Policy, set Provider to AWS, enter your 12-digit AWS account ID, leave Role ARN blank to trust any role in the account, set a max session duration, and add the required permissions to the allowlist.
    Pin a trust policy to a single Lambda function's execution role so only that workload can mint a StepSecurity API token.
    📍In the StepSecurity console, go to Organization › Trust Policies › New Policy, set Provider to AWS, enter the 12-digit AWS account ID, and enter the full role ARN (e.g. arn:aws:iam::123456789012:role/my-lambda-role) to restrict minting to that single workload.
    • Adds AWS as a second OIDC federation provider for StepSecurity trust policies, alongside 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. — EKS, EC2, Lambda, and any workload that can call AWS STSAWS STSAWS Security Token Service, an Amazon Web service that issues short-lived credentials for IAM roles, letting tools authenticate to AWS APIs without storing long-term secrets. can now mint scoped StepSecurity API tokens without a long-lived key.
    • AWS trust policies match on the issuer URL of the account's outbound federation, the 12-digit AWS account ID, and an optional role ARN — leave the role ARN blank to trust any role in the account, or specify one to pin to a single workload.
    • AWS trust policies carry the same max session duration and permissions allowlist controls already available to GitHub Actions policies, so minted tokens are capped and scoped identically.
    • AWS provider is available at organization scope via the StepSecurity API (Org Access) trust policy form, which now opens with a provider selector — either GitHub Actions or AWS — and shows the matching identity fields for whichever provider is chosen.

    4 details in Everything

33. See Which Threat Center Incidents Affect You (2026-08-18) Aug 18, 2026 · issue 049

    StepSecurity Threat Center now evaluates incidents against your environment and surfaces only the ones that affect you.

    ⤢ ⤢
    • Adds an Affecting You filter at the top of Threat Center to switch between All incidents and a scoped view showing only incidents that match your environment, each with a live count.
    • Adds a You are affected badge to incidents in both the All and Affecting You views so exposure is visible directly from the list.
    • New matched-components breakdown per affected incident lists compromised packages found in your repositories, split by open pull request vs. default branch, down to the affected files, head commit, and matched version.
    • New 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. correlation maps indicators of compromiseindicators of compromiseAn 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. (such as malicious network endpoints) from an incident against Harden-Runner runtime data, with a per-run breakdown of repository, workflow, job, and run.
    • Adds Affected packages and Exact version only notification granularity options so alerts fire only for incidents that touch your own dependencies.

    5 details in Everything

34. Harden Runner Releases: August 17, 2026 (2026-08-17) Aug 17, 2026 · issue 049

    Harden Runner gains a deny list egress policy via denied-endpoints and denied_endpoints, plus AWS CodeBuild custom image support.

    ⤢ ⤢
    └──▷ USE IT
    Block traffic to a known-bad domain and a wildcard subdomain range in a 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. workflow without allowlisting every other destination.
    .github/workflows/ci.yml copy

    - uses: step-security/harden-runner@v2.21.0
      with:
        egress-policy: block
        denied-endpoints: |
          malicious.example.com
          *.ads.example.org

    • Adds denied-endpoints input to the Harden-Runner GitHub Action (v2.21.0) to block specific destinations while leaving all other traffic reachable — use with egress-policy: block to stop known-bad hosts without building a full allow list.
    • Adds denied_endpoints config key to Self-Hosted VM Harden Runner (v1.9.0) with egress_policy: block for the same deny list capability on self-hosted VMs.
    • Deny lists for both the GitHub Action and Self-Hosted VM runner can be managed centrally through the Policy Store, eliminating the need to edit workflow or config files per job.
    • Extends support to AWS CodeBuild runners using Linux containers on EC2EC2Amazon's cloud virtual-machine service, letting tools run on-demand servers without managing physical hardware. compute, covering the custom image scenario.

    4 details in Everything

35. Harden Runner Releases: August 17, 2026 Aug 17, 2026 · release notes

    Harden Runner adds deny list egress policy via new inputs and Policy Store support

    └──▷ HOW TO FIND IT
    Block specific destinations while allowing all other egress traffic using workflow inputs
    📍egress-policy: block denied-endpoints: | example.com *.example.org
    • Adds egress-policy: block and denied-endpoints inputs to the Harden-Runner GitHub Action to explicitly block traffic to specific destinations
    • Enables management of deny lists through the central Policy Store instead of editing individual workflow files

    2 details in Everything

36. Secure Registry Now Supports Go (2026-08-13) Aug 13, 2026 · issue 045

    StepSecurity Secure Registry adds Go module proxying in Beta, with cooldown periods, block lists, and tag-only enforcement.

    ⤢ ⤢
    • Adds Go module proxy support to Secure Registry (Beta) at https://registry.stepsecurity.io/go, joining npm, PyPI, Maven, and NuGetNuGetMicrosoft's package manager for .NET, hosting a public registry of reusable libraries so tools can pull in dependencies without bundling or manually maintaining third-party code.; every module request is evaluated against your Go policy before a response is returned.
    • New Released Tags Only control (Go-specific) rejects module resolutions at commit hashes, branch names, and pseudo-versions such as v0.0.0-20241114170450-2d3c2a9cc518; only published release tags are permitted.
    • Cooldown Period holds newly published Go module versions for a configurable window, with per-module exemptions scoped to the full module path.
    • Compromised Packages control blocks Go module versions flagged by StepSecurity's 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. or the wider security community.
    • Custom Block List bans a Go module or version range org-wide using module@version form with glob support.

    +3 more show less

    • New Proxy Checksum Database setting under Go Registry Settings relays sum.golang.org requests through Secure Registry, removing the need to disable GOSUMDB when runners cannot reach the default checksum database.
    • Every Go module request is logged in the Policy Evaluations log as Allowed, Modified, or Blocked, attributed to the developer machine or workflow run that made the request.
    • Integration paths for Go proxy support cover JFrog Artifactory, Sonatype Nexus Repository, and Direct.

    8 details in Everything

37. Self-Hosted Runners: Full Fleet Visibility and a New GitHub App Permission (2026-08-11) Aug 11, 2026 · issue 043

    StepSecurity Self-Hosted Runners page now shows a unified fleet inventory including unmonitored runners, requiring a new GitHub App permission.

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Approve the new GitHub App permission so your full runner fleet — including organization-level runners never seen by the agent — appears in the StepSecurity inventory.
    📍In GitHub, go to Organization Settings → GitHub Apps, locate the StepSecurity app, and approve the pending Self-hosted runners: Read-only permission request.
    • New Self-hosted runners: Read-only organization permission added to the StepSecurity GitHub App enables listing runners registered at the organization level via the GitHub API, not just those reporting from the Harden-Runner agent.
    • Admins must approve the pending GitHub App permission request under Organization Settings → GitHub Apps for organization-level runners to appear in the inventory.
    • The Self-Hosted Runners page now merges runners discovered from GitHub (org-level and repo-level) with agent-reported runners into a single consolidated inventory, surfacing runners with no Harden-Runner agent installed as Not installed in the Harden Runner column.
    • Each runner entry now shows a discovery source, registration scope badge (Org or Repo with repository name), and lifecycle classification (Persistent or Ephemeral).
    • New per-runner detail panel exposes agent version and install time for agent-reported runners, plus level, OS, runner group, labels, and observed job counts for GitHub-registered runners.

    +1 more show less

    • Fleet search and filter capabilities added: search by runner name or label, filter by status, view last-sync timestamp, and trigger an on-demand refresh.

    6 details in Everything

38. Harden Runner Releases- August 5, 2026 (2026-08-05) Aug 5, 2026 · issue 036

    Harden Runner adds AWS CodeBuild runner support, container process tracing, versioned job hooks v1.0.0, and implicit single-label domain allowances in block mode.

    ⤢ ⤢
    └──▷ USE IT
    Enable container process tracing on ARC-hosted pods to capture process events from inside containers.
    yaml copy

    env:
      enableContainerProcessTracing: "true"

    Configure versioned Runner Job Hooks on self-hosted runners via the StepSecurity app UI.
    📍In the StepSecurity app, go to Settings → Harden-Runner Installation → Runner Job Hooks and follow the installation instructions to reference a specific version from step-security/gha-runner-job-hooks.
    • Adds support for 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. runners hosted on AWS CodeBuild in Harden-Runner GitHub Action v2.20.1.
    • Adds container process tracing (opt-in) for ARC pods — enable via Helm value env: enableContainerProcessTracing: "true" in ARC Harden Runner v2.18.6.
    • Single-labeled internal hostnames (no dot) are now implicitly allowed when egress-policy: block is set across all three deployment options (GitHub Action v2.20.1, ARC v2.18.6, Self-Hosted VM v1.8.14), removing the need to list them in allowed-endpoints.
    • Publishes Runner Job Hooks v1.0.0 at step-security/gha-runner-job-hooks, enabling runners to pin to a specific hook version for pre-job and post-job execution.
    • Pre-job and post-job hooks rewritten from shell script to TypeScript/JavaScript in Runner Job Hooks v1.0.0.

    5 details in Everything

39. Harden Runner Releases- August 5, 2026 Aug 5, 2026 · release notes

    Harden Runner introduces Runner Job Hooks v1.0.0 with versioned TypeScript hooks and ARC container process tracing

    • Adds Runner Job Hooks feature to allow referencing specific hook versions via step-security/gha-runner-job-hooks
    • Enables opt-in container process tracing for ARC Harden Runner v2.18.6 via Helm values
    • Implicitly allows single-labeled internal domains in block mode across all Harden Runner deployment options

    3 details in Everything

40. Threat Intel Notifications: Tenant-Wide Channels and Match Granularity (2026-08-04) Aug 4, 2026 · issue 036

    StepSecurity threat intel notifications gain three match-granularity levels and a tenant-wide channel configuration page.

    ⤢ ⤢
    └──▷ HOW TO FIND IT
    Configure a single Slack webhook for all organizations in your tenant so every team receives threat intel alerts without repeating channel setup per org.
    📍In the Admin console, go to Settings › Notifications and enter your Slack webhook URL under the Slack webhook field, then save.
    Limit threat intel noise for an application team by alerting only when the exact compromised package version is present in their environment.
    📍In organization notification settings, enable the 'Notify when StepSecurity threat intel flags a compromised component' event, open the Threat intel notifications dialog, and select Exact version only.
    • Adds a Notifications page under Admin console settings to configure email, Slack webhook, and Teams webhook once at the tenant level, replacing per-organization channel setup.
    • Introduces three match-granularity levels for threat intel notifications — All threat intel incidents, Affected packages, and Exact version only — selectable from the Threat intel notifications dialog.
    • Makes threat intel alerting an explicit opt-in per organization via the Notify when StepSecurity threat intel flags a compromised component event in organization notification settings, shown alongside existing Harden-Runner detection events.
    • Allows granularity to be set independently at the tenant level so a tenant can subscribe to the full incident feed regardless of individual organization configuration.

    4 details in Everything

41. Dev Machine Guard v1.15.0 (2026-08-03) Aug 3, 2026 · issue 035

    Dev Machine Guard v1.15.0 adds npm/pnpm/Yarn/bun registry enforcement, VS Code private marketplace URLs, and an MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies. verify-only channel.

    ⤢ ⤢
    • Adds package_config policy support: the agent converges a StepSecurity-owned block in ~/.npmrc to route npm, pnpm, Yarn classic, and bun through your tenant's Secure Registry; absent policy is a no-op with snapshot rollback.
    • Writes extensions.gallery.serviceUrl into VS Code user-scope settings.json atomically alongside extensions.allowed when a policy sets a gallery URL; removal is ownership-gated so developer-configured URLs are never deleted.
    • Introduces a verify-only enforcement channel for MDM-enforced profiles: reads the OS-managed VS Code policy (Windows registry, macOS managed preferences, /etc/vscode/policy.json) or effective ~/.npmrc and reports observed values and drift without writing.
    • Records device policy ownership for every category in a single locked state file, preventing one category's policy clear from dropping another category's ownership record.
    • Detects Homebrew installed outside PATH by resolving brew from /opt/homebrew, /usr/local, and /home/linuxbrew and reading version metadata from Homebrew's repository instead of invoking brew --version, completing formula and cask inventory on shell-less scans.

    +1 more show less

    • Makes VS Code settings.json files beginning with a UTF-8 byte order mark (written by PowerShell 5.1 Set-Content -Encoding UTF8 or utf8bom-configured editors) enforceable.

    6 details in Everything

42. Device Policy: Package Config Policies, MDM Verification, and Private Marketplace (2026-08-03) Aug 3, 2026 · issue 035

    Device Policy adds npm package config policies, MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies. drift verification, private marketplace support, and per-profile enforcement types.

    ⤢ ⤢
    • Adds npm package config policies that write a managed block into ~/.npmrc with the tenant Secure Registry URL and a per-device ::dev: suffixed authentication token, bounded by BEGIN/END markers to leave surrounding configuration untouched.
    • Adds MDM verification with field-by-field drift diffs: the agent continuously compares what your MDM deployed against the policy and reports one of eight states (MDM verified, MDM drift, Not deployed, Verification failed, and others), showing desired vs. observed for every field.
    • Generates audit, remediation, and uninstall script sets per platform for MDM-delivered package config policies, designed for deployment into Intune Remediations detection/remediation slots or an Iru Custom Script library.
    • Adds an optional gallery URL on IDE extension policies that compiles to extensions.gallery.serviceUrl, pointing assigned devices at a self-hosted extension gallery instead of the public Visual Studio Marketplace; the gallery URL is itself verified for drift.
    • Every profile now declares an explicit enforcement type — Verify only (MDM-enforced) or Self healing (agent-enforced) — and switching an existing profile between types migrates all assigned devices, which report Pending until each agent checks in.

    5 details in Everything

43. NuGet Ecosystem Support in GitHub Checks, OSS Package Search, and Secure Registry (2026-07-30) Jul 30, 2026 · issue 035

    StepSecurity adds NuGetNuGetMicrosoft's package manager for .NET, hosting a public registry of reusable libraries so tools can pull in dependencies without bundling or manually maintaining third-party code. (.NET) ecosystem support to GitHub Checks, OSSOSSOpen-source software: code whose licence permits anyone to inspect, modify, and redistribute it. Cyber tools build on it to reuse auditable libraries without licensing fees. Package Search, and Secure Registry.

    ⤢ ⤢
    └──▷ TRY IT
    Perform a NuGet restore without triggering NU1905 warnings while NuGetAudit support is pending in Secure Registry.
    copy

    $ dotnet restore -p:NuGetAudit=false

    • Adds NuGet Package Compromised Updates PR Content Security control to GitHub Checks, blocking pull requests that introduce or update a NuGet dependency flagged in StepSecurity's compromised-packages database; can be set to Required or Optional.
    • Adds NuGet Package Cooldown PR Content Security control to GitHub Checks, blocking pull requests that add recently published NuGet versions until they age beyond the configured cooldown window; can be set to Required or Optional.
    • Secure Registry now serves NuGet (Beta) at https://registry.stepsecurity.io/dotnet/index.json (V3 protocol only; resources under /dotnet/v3/), routing all NuGet metadata requests and package downloads through Secure Registry for policy evaluation.
    • Secure Registry NuGet policy controls available at launch: Cooldown Period, Compromised Packages, and Custom Block List; TyposquattingTyposquattingUsing a name close to a trusted domain or software package to catch spelling mistakes. Attackers use it to lure people or installers to a malicious lookalike. Protection is marked Coming Soon.
    • Secure Registry supports three NuGet integration paths: JFrog Artifactory, Sonatype Nexus Repository, and Direct; the Direct path includes a Package manager selector that switches setup instructions between the .NET CLI, the NuGet CLI, and Visual Studio.

    +4 more show less

    • NuGet is now selectable from the Package ecosystem toggle in OSS Package Search; packages are indexed from *.csproj, *.vbproj, *.fsproj, packages.config, packages.lock.json, Directory.Packages.props, Directory.Build.props, and *.deps.json files.
    • Policy Evaluations audit log now includes .NET in the Ecosystem filter, enabling narrowing of audit log entries to NuGet requests.
    • NuGet joins the Registry selector on both the Policy and Setup Guide tabs in Secure Registry, with policy configured independently per ecosystem.
    • NuGetAudit is not yet supported: Secure Registry does not proxy the VulnerabilityInfo resource, causing audit-enabled restores to emit NU1905; workaround is to pass -p:NuGetAudit=false or exclude the Secure Registry source from audit sources.

    9 details in Everything

44. Secure Registry: Maven Support in Beta, Custom Block List, and Tarball URL Rewriting (2026-07-24) Jul 24, 2026 · issue 028

    StepSecurity Secure Registry adds Maven (Beta) support, per-ecosystem Custom Block List, and npm Tarball URL Rewriting.

    └──▷ HOW TO FIND IT
    Enforce an internal ban on all 4.x versions of lodash and an entire scoped namespace across your npm policy.
    📍In the Secure Registry console, go to Policy › npm › Custom Block List and add entries: lodash@4* and @rejected-scope/*, then save.
    Route all Maven artifact downloads through Secure Registry for evaluation instead of Maven Central.
    📍In the Secure Registry console, go to Setup Guide › Registry selector › Maven, select your integration path (e.g., Direct), then point your Maven client at https://registry.stepsecurity.io/java as the mirror URL.
    Ensure npm package tarballs download through Secure Registry rather than the public npm registry.
    📍In the Secure Registry console, go to npm Registry Settings and enable the Tarball URL Rewriting toggle.
    • Adds Maven (Java) ecosystem support (Beta) to Secure Registry — route all Maven metadata and artifact downloads through https://registry.stepsecurity.io/java with Cooldown Period, Compromised Packages, and Custom Block List controls at launch.
    • Supports four Maven integration paths: JFrog Artifactory, Google Artifact Registry, Sonatype Nexus Repository, and Direct, with setup steps switchable via a new Package manager selector on the Setup Guide tab.
    • Introduces Custom Block List, a new per-ecosystem control that blocks specific packages or versions using glob patterns (e.g., lodash@4*, @scope/*), evaluated independently of Cooldown, Compromised Packages, and TyposquattingTyposquattingUsing a name close to a trusted domain or software package to catch spelling mistakes. Attackers use it to lure people or installers to a malicious lookalike. Protection.
    • Adds Tarball URL Rewriting setting for npm, rewriting tarball URLs in metadata responses so package downloads route through Secure Registry instead of the public npm registry.

    4 details in Everything

45. docs update Jul 22, 2026 · issue 022

    StepSecurity adds Agent Skills support for developer machines.

    • New Agent Skills capability for developer machines enables automated security skill execution in agentic workflows.

    1 detail in Everything

46. docs update Jul 22, 2026 · docs issue 004

    StepSecurity adds Agent Skills support for developer machines.

    • New Agent Skills capability for developer machines enables automated security skill execution in agentic workflows.

    1 detail in Everything

47. Workflow Run Policies: allow-list enforcement, runner constraints, and richer policy insight (2026-07-21) Jul 21, 2026 · issue 022

    StepSecurity Workflow Run Policies gain allow-list enforcement, container blocking, bulk-secret detection, and richer per-run evaluation visibility.

    └──▷ HOW TO FIND IT
    Block bulk secret harvesting (e.g. toJSON(secrets)) across all branches while allowing normal targeted secret access — configure in the Secret Exfiltration Policy.
    📍In the StepSecurity console, go to Workflow Run Policies › Secret Exfiltration Policy, enable 'Block only when a workflow accesses all secrets at once' (recommended), then enable Analyze default branch runs to extend coverage to the default branch.
    Restrict CI jobs to an approved set of runner labels and RunsOn structured constraints, preventing use of arbitrary or untrusted runner configurations.
    📍In the StepSecurity console, go to Workflow Run Policies › Runner Label Policy and switch to Allowed labels & constraints mode. Add the permitted labels and structured runs-on keys (e.g. family, cpu, image) to the allow list; each configured dimension is enforced independently.
    Ensure every targeted job sources its runtime policy from the StepSecurity policy store, preventing jobs from bypassing centrally managed policies.
    📍In the StepSecurity console, go to Workflow Run Policies › Harden-Runner Policy and enable Require policy store. Targeted jobs must then include use-policy-store: true on their Harden-Runner step or the run will be blocked.
    • Adds Currently known compromised actions display in the Compromised Actions Policy editor, showing affected versions, with the list snapshotted per evaluation for auditability.
    • New 'Block only when a workflow accesses all secrets at once' mode in Secret Exfiltration Policy blocks bulk patterns like ${{ toJSON(secrets) }} while allowing targeted references, reducing false positives.
    • New Analyze default branch runs option in Secret Exfiltration Policy extends coverage to the default branch.
    • Secret Exfiltration violation details now include the specific secret reference and its line number in the workflow file.
    • New Allowed labels & constraints allow-list mode in Runner Label Policy enforces that only configured labels and constraint values may be used, with each dimension enforced independently.

    +7 more show less

    • Runner Label Policy now supports structured runs-on keys (family, cpu, image, ami, volume, etc.) for third-party runner providers such as RunsOn.
    • New Require policy store option in Harden-Runner Policy enforces that targeted jobs set use-policy-store: true on their Harden-Runner step.
    • New Block jobs running in a container option in Harden-Runner Policy blocks jobs with a job-level container: on GitHub-hosted standard runners where outbound monitoring is unavailable.
    • Adds customizable PR Comment Template per policy with placeholders including {{policy_name}}, {{policy_details}}, {{remediation}}, and {{docs_url}}.
    • Policy Evaluations page now shows evaluation status (Allowed, DryRunBlocked, or Blocked), per-policy violation breakdown, related PR link, and run policy timing metrics from webhook receipt through cancellation.
    • Workflow run insights now include a Workflow Run Policy Evaluation section on the Summary tab covering blocked, allowed, and no-policy states, including for cancelled runs.
    • Network Events tab in Harden-Runner insights now shows the number of suppression rules applied per job and a View suppression rules link for auditing missed detections.

    12 details in Everything

48. Agent Skills Inventory for Developer Machines (2026-07-17) Jul 17, 2026 · issue 022

    Dev Machine Guard adds fleet-wide AI agent skills inventory with executable content flagging and provenanceprovenanceEvidence recording where a software artifact came from and how it was built, including its source and build process. It helps a consumer check that the artifact matches a trusted build. tracking.

    └──▷ HOW TO FIND IT
    Review which AI agent skills on your fleet ship executable content or lifecycle hooks before they run on a developer's machine.
    📍In the Dev Machine Guard console, go to Agent Skills, then filter by Flags → hooks or shell to see every skill that registers lifecycle hooks or injects shell commands, and drill into per-device installation details.
    Spot developer machines running locally edited or outdated skill copies that differ from the rest of the fleet.
    📍In the Dev Machine Guard console, go to Agent Skills and look for skills highlighted under version drift detection (SKILL.md content hash mismatch), then use Export CSV to capture the full list for remediation tracking.
    • New Agent Skills page in the Dev Machine Guard dashboard inventories every AI agent skill (Claude Code, Codex, GitHub Copilot) detected across your fleet, with per-device installation details and scope (Global, Project, or System).
    • Flags skills that contain executable script files, declare lifecycle hooks, or inject shell commands that run with developer privileges — surfacing supply chain risk from AI coding agent skill folders.
    • Traces provenance of skills installed via the skills.sh CLI back to their source GitHub repository, plugin, and upstream folder hash; standalone skills are marked as authored on device.
    • Detects version drift by comparing SKILL.md content hashes across devices and highlights skills that differ between machines.
    • Supports filtering by source (skills.sh managed or local), agent, scope, and executable-content flags, plus full inventory export as CSV.

    5 details in Everything

49. New Control: Jobs Should Use a Secure Registry Instead of Public Package Registries (2026-07-14) Jul 14, 2026 · issue 017

    StepSecurity adds a secure-registry CI/CD control and OIDCOIDCOpenID Connect, a sign-in protocol built on OAuth 2.0. An application uses identity information from a trusted provider instead of maintaining a separate password for each user. replaceability tracking for Actions secrets.

    └──▷ HOW TO FIND IT
    Identify CI/CD jobs pulling packages from public registries so you can route them through a secure proxy before a supply-chain compromise occurs.
    📍In the StepSecurity console, go to Overview dashboard › Controls, open the Jobs Should Use a Secure Registry Instead of Public Package Registries control, then filter findings by repository or workflow to prioritize remediation.
    Find secrets that can be eliminated entirely by switching to OIDC, reducing long-lived credential exposure across your repositories.
    📍In the StepSecurity console, go to Actions Secret › Repository Secrets tab, then enable the OIDC Replaceable filter to list only secrets (e.g., NPM_TOKEN) that trusted publishing can replace.
    • New Overview dashboard control detects CI/CD jobs that download packages directly from public registries (registry.npmjs.org, pypi.org, registry-1.docker.io) instead of a secure proxy registry.
    • Control findings are filterable by repository, workflow, and compliance status, with remediation tracking and finding suppression on the control detail page.
    • Repository Secrets tab now surfaces Usage Status (Active, Stale, Unused, Unknown) for each secret, including which workflow file and step references it and the last run time and conclusion.
    • Secrets that can be replaced with OIDC (e.g., NPM_TOKEN replaceable via npm trusted publishing) are flagged in the Actions Secret section.
    • New summary cards on the Repository Secrets tab show Total Secrets, Unused Secrets, Stale Secrets, and OIDC Replaceable counts for an at-a-glance secrets posture view.

    +1 more show less

    • Filtering on the Repository Secrets tab supports search by repository or secret name, status filter, and an OIDC-replaceable-only toggle.

    6 details in Everything

50. Dev Machine Guard v1.13.0 (2026-07-09) Jul 9, 2026 · issue 014

    Dev Machine Guard v1.13.0 makes Device Policy enforcement GA and adds Visual Studio discovery and disk-based package scanning.

    └──▷ HOW TO FIND IT
    Assign a Device Policy profile to enforce an approved VS Code extension set across all developer machines without MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies..
    📍In the dashboard, go to Device Policy › Profiles, create or select a profile, then assign it to All Devices. The agent installs and continuously enforces the approved extension set on its next cycle.
    • Device Policy enforcement via the DMG agent is now generally available — enforces approved VS Code extension sets on developer machines without requiring MDM, assignable per device or globally from the dashboard.
    • New classic Visual Studio installation discovery extends IDE extension inventory beyond VS Code to legacy Visual Studio versions and their extensions.
    • New disk-based package scanning for npm (lockfile parsing) and Python (dist-info metadata) with a scan cache and concurrent workers for faster runs; previous method remains available via the use_legacy_package_scan opt-out.
    • Scans can now be triggered at login, with improved scheduling reliability and scheduler diagnostics.
    • Expanded global package discovery adds support for pnpm v10+ globals and Yarn classic globals on Windows.

    5 details in Everything

51. Dev Machine Guard v1.13.0 (2026-07-09) Jul 9, 2026 · issue 002

    Dev Machine Guard v1.13.0 makes Device Policy enforcement GA and adds Visual Studio discovery and disk-based package scanning.

    └──▷ HOW TO FIND IT
    Assign a Device Policy profile to enforce an approved VS Code extension set across all developer machines without MDMMDMMobile device management, centrally configuring and managing enrolled devices. Administrators use it to apply settings, distribute applications and enforce device security policies..
    📍In the dashboard, go to Device Policy › Profiles, create or select a profile, then assign it to All Devices. The agent installs and continuously enforces the approved extension set on its next cycle.
    • Device Policy enforcement via the DMG agent is now generally available — enforces approved VS Code extension sets on developer machines without requiring MDM, assignable per device or globally from the dashboard.
    • New classic Visual Studio installation discovery extends IDE extension inventory beyond VS Code to legacy Visual Studio versions and their extensions.
    • New disk-based package scanning for npm (lockfile parsing) and Python (dist-info metadata) with a scan cache and concurrent workers for faster runs; previous method remains available via the use_legacy_package_scan opt-out.
    • Scans can now be triggered at login, with improved scheduling reliability and scheduler diagnostics.
    • Expanded global package discovery adds support for pnpm v10+ globals and Yarn classic globals on Windows.

    5 details in Everything

52. Harden Runner Releases- July 7, 2026 (2026-07-07) Jul 7, 2026 · issue 014

    Harden Runner adds egress block mode for macOS/Windows, Bitrise support, and Bun HTTPSHTTPSHTTP carried over an encrypted TLS connection. It protects traffic in transit and lets a browser verify the server's certificate. monitoring.

    └──▷ USE IT
    Restrict outbound traffic on a Windows GitHub-hosted runner to only allowed endpoints — now that block mode is supported beyond Linux.
    yaml copy

    - uses: step-security/harden-runner@v2.20.0
      with:
        egress-policy: block
        allowed-endpoints: >
          api.nuget.org:443
          github.com:443

    • Supports egress-policy: block mode on macOS and Windows GitHub-hosted runners, enabling outbound traffic restriction (not just auditing) on these platforms.
    • Adds compatibility with Bitrise-hosted macOS runners for 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. workflows.
    • Enables HTTPS request monitoring for workflows using the Bun runtime on Linux runners (enterprise tier) in the GitHub Action.
    • Enables HTTPS request monitoring for Bun on Linux amd64 self-hosted VM runners.

    4 details in Everything

53. Harden Runner Releases- July 7, 2026 (2026-07-07) Jul 7, 2026 · issue 002

    Harden Runner adds egress block mode for macOS/Windows, Bitrise support, and Bun HTTPSHTTPSHTTP carried over an encrypted TLS connection. It protects traffic in transit and lets a browser verify the server's certificate. monitoring.

    └──▷ USE IT
    Restrict outbound traffic on a Windows GitHub-hosted runner to only allowed endpoints — now that block mode is supported beyond Linux.
    yaml copy

    - uses: step-security/harden-runner@v2.20.0
      with:
        egress-policy: block
        allowed-endpoints: >
          api.nuget.org:443
          github.com:443

    • Supports egress-policy: block mode on macOS and Windows GitHub-hosted runners, enabling outbound traffic restriction (not just auditing) on these platforms.
    • Adds compatibility with Bitrise-hosted macOS runners for 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. workflows.
    • Enables HTTPS request monitoring for workflows using the Bun runtime on Linux runners (enterprise tier) in the GitHub Action.
    • Enables HTTPS request monitoring for Bun on Linux amd64 self-hosted VM runners.

    4 details in Everything

54. Harden Runner Releases- July 7, 2026 Jul 7, 2026 · release notes

    Harden Runner v2.20.0 adds egress block policy on macOS/Windows runners and HTTPSHTTPSHTTP carried over an encrypted TLS connection. It protects traffic in transit and lets a browser verify the server's certificate. monitoring for Bun runtime

    • Adds egress-policy: block mode support for macOS and Windows GitHub-hosted runners to restrict outbound traffic
    • Adds HTTPS request monitoring support for the Bun runtime on Linux GitHub-hosted runners (enterprise tier)
    • Adds support for Bitrise-hosted macOS 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. runners

    3 details in Everything

55. New Detection: Secret Exfiltration Attempt (2026-06-30) Jun 30, 2026 · issue 014

    Harden-Runner adds Secret Exfiltration Attempt detection to catch toJSON(secrets) supply-chain attacks in real time.

    └──▷ HOW TO FIND IT
    Get an immediate alert when a secret exfiltration attempt is detected, so you can rotate exposed tokens before an attacker retrieves the artifact.
    📍In the Harden-Runner console, go to Notification Settings and enable the Secret Exfiltration Attempt detection notification.
    • New Secret Exfiltration Attempt detection type flags workflow runs that combine known attack indicators — malicious workflow name, toJSON(secrets) secret dump, suspicious commit message, and an exfiltration artifact — before stolen secrets can be retrieved.
    • Confidence scoring elevates a detection to High when the workflow run, branch, or workflow file is deleted shortly after execution, treating evidence removal as a strong exploitation signal.
    • Direct navigation in the Detections page pinpoints the affected repository, matched rule, indicators, workflow file, and the exact job, step, and line where the secret dump occurred.
    • Opt-in alerting available in Notification Settings to receive an immediate notification the moment a Secret Exfiltration Attempt is flagged.

    4 details in Everything

56. New Detection: Secret Exfiltration Attempt (2026-06-30) Jun 30, 2026 · issue 002

    Harden-Runner adds Secret Exfiltration Attempt detection to catch toJSON(secrets) supply-chain attacks in real time.

    └──▷ HOW TO FIND IT
    Get an immediate alert when a secret exfiltration attempt is detected, so you can rotate exposed tokens before an attacker retrieves the artifact.
    📍In the Harden-Runner console, go to Notification Settings and enable the Secret Exfiltration Attempt detection notification.
    • New Secret Exfiltration Attempt detection type flags workflow runs that combine known attack indicators — malicious workflow name, toJSON(secrets) secret dump, suspicious commit message, and an exfiltration artifact — before stolen secrets can be retrieved.
    • Confidence scoring elevates a detection to High when the workflow run, branch, or workflow file is deleted shortly after execution, treating evidence removal as a strong exploitation signal.
    • Direct navigation in the Detections page pinpoints the affected repository, matched rule, indicators, workflow file, and the exact job, step, and line where the secret dump occurred.
    • Opt-in alerting available in Notification Settings to receive an immediate notification the moment a Secret Exfiltration Attempt is flagged.

    4 details in Everything

57. StepSecurity Maintained Actions Are Now Free for Public Repositories (2026-06-29) Jun 29, 2026 · issue 014

    StepSecurity Maintained Actions — secure drop-in replacements for 500+ 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. — are now free for public repositories.

58. StepSecurity Maintained Actions Are Now Free for Public Repositories (2026-06-29) Jun 29, 2026 · issue 002

    StepSecurity Maintained Actions — secure drop-in replacements for 500+ 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. — are now free for public repositories.

59. Harden Runner Releases- June 24, 2026 (2026-06-24) Jun 24, 2026 · issue 014

    ARC Harden Runner v2.18.3 adds broker pod customization fields and Cosign keyless image signing.

    └──▷ TRY IT
    Verify the arc-harden-runner container image's provenanceprovenanceEvidence recording where a software artifact came from and how it was built, including its source and build process. It helps a consumer check that the artifact matches a trusted build. before deploying, ensuring it was built by the official 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. pipeline.
    copy

    $ cosign verify ghcr.io/step-security/arc-harden-runner/arc-harden-runner:v2.18.3 \
      --certificate-identity-regexp 'https://github.com/step-security/arc-harden-runner/.*' \
      --certificate-oidc-issuer https://token.actions.githubusercontent.com

    Verify the broker sidecar image's provenance to confirm supply-chain integrity before cluster deployment.
    copy

    $ cosign verify ghcr.io/step-security/broker/broker:v1.3.0 \
      --certificate-identity-regexp 'https://github.com/step-security/broker/.*' \
      --certificate-oidc-issuer https://token.actions.githubusercontent.com

    • Adds broker pod-spec customization via new Helm values under broker.* (volumeMounts, imagePullSecrets, automountServiceAccountToken, serviceAccountName, resources, nodeSelector, tolerations, affinity, topologySpreadConstraints) for scheduling alongside strict workloads.
    • Supports Cosign keyless (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.) signing for versioned and FIPSFIPSA set of US federal standards published by NIST that define approved cryptographic algorithms and modules; cyber tools use it to meet compliance requirements in government and regulated environments. arc-harden-runner and broker image manifests published to ghcr.io, enabling supply-chain verification.

    2 details in Everything

60. Secure Registry: Source Attribution, Typosquatting Protection, and Sonatype Nexus Support (2026-06-24) Jun 24, 2026 · issue 014

    Secure Registry gains source attribution, typosquattingtyposquattingUsing a name close to a trusted domain or software package to catch spelling mistakes. Attackers use it to lure people or installers to a malicious lookalike. blocking, and Sonatype Nexus Repository support.

    └──▷ HOW TO FIND IT
    Attribute CI pipeline package requests to a specific 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. run so Policy Evaluations log entries link back to the correct workflow.
    📍Set your Secure Registry API key token to the form <token>::gha:<identifier> in your pipeline's package manager configuration, following the Setup Guide.
    Route npm traffic through Sonatype Nexus Repository while enforcing Secure Registry policy, using the new Nexus integration path.
    📍In the Secure Registry Setup Guide, select Sonatype Nexus Repository as the integration path and configure Nexus to proxy npm through your Secure Registry endpoint.
    Enable Typosquatting Protection for an ecosystem and allowlist a legitimate package that triggers a false positive.
    📍In Secure Registry settings, enable Typosquatting Protection for the target ecosystem, then add the exact package name to the Whitelisted Packages allowlist to permit it.
    • Adds Source Attribution to the Policy Evaluations log via two new columns — Source and Source Identifier — tracing every package request to the developer machine or CI pipeline that made it.
    • Enables attribution opt-in by appending an identifier suffix to your API key token (::dev: for developer machines, gha: for GitHub Actions pipeline runs).
    • New Typosquatting Protection control blocks packages whose names closely resemble popular packages, configurable per ecosystem with a Whitelisted Packages allowlist to clear false positives by exact package name.
    • Adds Sonatype Nexus Repository as an integration path for teams that proxy npm through Nexus, joining JFrog Artifactory, Google Artifact Registry, and Direct npm.

    4 details in Everything

61. Secure Registry: Source Attribution, Typosquatting Protection, and Sonatype Nexus Support (2026-06-24) Jun 24, 2026 · issue 002

    Secure Registry gains source attribution, typosquattingtyposquattingUsing a name close to a trusted domain or software package to catch spelling mistakes. Attackers use it to lure people or installers to a malicious lookalike. blocking, and Sonatype Nexus Repository support.

    └──▷ HOW TO FIND IT
    Attribute CI pipeline package requests to a specific 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. run so Policy Evaluations log entries link back to the correct workflow.
    📍Set your Secure Registry API key token to the form <token>::gha:<identifier> in your pipeline's package manager configuration, following the Setup Guide.
    Route npm traffic through Sonatype Nexus Repository while enforcing Secure Registry policy, using the new Nexus integration path.
    📍In the Secure Registry Setup Guide, select Sonatype Nexus Repository as the integration path and configure Nexus to proxy npm through your Secure Registry endpoint.
    Enable Typosquatting Protection for an ecosystem and allowlist a legitimate package that triggers a false positive.
    📍In Secure Registry settings, enable Typosquatting Protection for the target ecosystem, then add the exact package name to the Whitelisted Packages allowlist to permit it.
    • Adds Source Attribution to the Policy Evaluations log via two new columns — Source and Source Identifier — tracing every package request to the developer machine or CI pipeline that made it.
    • Enables attribution opt-in by appending an identifier suffix to your API key token (::dev: for developer machines, gha: for GitHub Actions pipeline runs).
    • New Typosquatting Protection control blocks packages whose names closely resemble popular packages, configurable per ecosystem with a Whitelisted Packages allowlist to clear false positives by exact package name.
    • Adds Sonatype Nexus Repository as an integration path for teams that proxy npm through Nexus, joining JFrog Artifactory, Google Artifact Registry, and Direct npm.

    4 details in Everything

62. Harden Runner Releases- June 24, 2026 (2026-06-24) Jun 24, 2026 · issue 002

    ARC Harden Runner v2.18.3 adds broker pod customization fields and Cosign keyless image signing.

    └──▷ TRY IT
    Verify the arc-harden-runner container image's provenanceprovenanceEvidence recording where a software artifact came from and how it was built, including its source and build process. It helps a consumer check that the artifact matches a trusted build. before deploying, ensuring it was built by the official 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. pipeline.
    copy

    $ cosign verify ghcr.io/step-security/arc-harden-runner/arc-harden-runner:v2.18.3 \
      --certificate-identity-regexp 'https://github.com/step-security/arc-harden-runner/.*' \
      --certificate-oidc-issuer https://token.actions.githubusercontent.com

    Verify the broker sidecar image's provenance to confirm supply-chain integrity before cluster deployment.
    copy

    $ cosign verify ghcr.io/step-security/broker/broker:v1.3.0 \
      --certificate-identity-regexp 'https://github.com/step-security/broker/.*' \
      --certificate-oidc-issuer https://token.actions.githubusercontent.com

    • Adds broker pod-spec customization via new Helm values under broker.* (volumeMounts, imagePullSecrets, automountServiceAccountToken, serviceAccountName, resources, nodeSelector, tolerations, affinity, topologySpreadConstraints) for scheduling alongside strict workloads.
    • Supports Cosign keyless (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.) signing for versioned and FIPSFIPSA set of US federal standards published by NIST that define approved cryptographic algorithms and modules; cyber tools use it to meet compliance requirements in government and regulated environments. arc-harden-runner and broker image manifests published to ghcr.io, enabling supply-chain verification.

    2 details in Everything

63. Harden Runner Releases- June 24, 2026 Jun 24, 2026 · release notes

    ARC Harden Runner v2.18.3 adds pod spec customization via broker.* Helm values and keyless Cosign image verification.

    └──▷ TRY IT
    Verify the signed arc-harden-runner image using cosign with 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. identity.
    copy

    $ cosign verify ghcr.io/step-security/arc-harden-runner/arc-harden-runner:v2.18.3 --certificate-identity-regexp 'https://github.com/step-security/arc-harden-runner/.*' --certificate-oidc-issuer https://token.actions.githubusercontent.com

    • broker.* Helm values (e.g., nodeSelector, tolerations) now supported for GitBook Assistant Broker pod customization
    • Arc-harden-runner v2.18.3 images are now signed with Cosign, verifiable via certificate identity and OIDC issuer

    2 details in Everything

64. Maven Ecosystem Support in GitHub Checks and OSS Package Search (2026-06-16) Jun 16, 2026 · issue 014

    StepSecurity adds Maven (Java) ecosystem support to GitHub Checks PR controls and OSSOSSOpen-source software: code whose licence permits anyone to inspect, modify, and redistribute it. Cyber tools build on it to reuse auditable libraries without licensing fees. Package Search.

    └──▷ HOW TO FIND IT
    Enable both new Maven PR controls to protect your Java repositories from compromised or freshly published dependencies.
    📍In the StepSecurity console, go to GitHub Checks › PR Content Security, select your Maven repository, and enable Maven Package Compromised Updates and Maven Package Cooldown — setting each to Required or Optional as your policy requires.
    Find where a specific Maven package appears across PRs, default branches, and developer machines to assess blast radius before incident response.
    📍In the StepSecurity console, go to OSS Package Search, select Maven from the Package ecosystem toggle, enter the package coordinates (e.g. com.fasterxml.jackson.core:jackson-databind), and review results across pull requests, default branches, and developer machines.
    • Adds Maven Package Compromised Updates PR control: blocks pull requests that introduce or update a Maven dependency flagged in StepSecurity's compromised-packages database.
    • Adds Maven Package Cooldown PR control: blocks pull requests adding recently published Maven versions until they age past the configured cooldown window.
    • Enables Maven ecosystem selection in OSS Package Search, allowing blast-radius analysis of specific Maven packages across pull requests, default branches, and developer machines.

    3 details in Everything

65. Maven Ecosystem Support in GitHub Checks and OSS Package Search (2026-06-16) Jun 16, 2026 · issue 002

    StepSecurity adds Maven (Java) ecosystem support to GitHub Checks PR controls and OSSOSSOpen-source software: code whose licence permits anyone to inspect, modify, and redistribute it. Cyber tools build on it to reuse auditable libraries without licensing fees. Package Search.

    └──▷ HOW TO FIND IT
    Enable both new Maven PR controls to protect your Java repositories from compromised or freshly published dependencies.
    📍In the StepSecurity console, go to GitHub Checks › PR Content Security, select your Maven repository, and enable Maven Package Compromised Updates and Maven Package Cooldown — setting each to Required or Optional as your policy requires.
    Find where a specific Maven package appears across PRs, default branches, and developer machines to assess blast radius before incident response.
    📍In the StepSecurity console, go to OSS Package Search, select Maven from the Package ecosystem toggle, enter the package coordinates (e.g. com.fasterxml.jackson.core:jackson-databind), and review results across pull requests, default branches, and developer machines.
    • Adds Maven Package Compromised Updates PR control: blocks pull requests that introduce or update a Maven dependency flagged in StepSecurity's compromised-packages database.
    • Adds Maven Package Cooldown PR control: blocks pull requests adding recently published Maven versions until they age past the configured cooldown window.
    • Enables Maven ecosystem selection in OSS Package Search, allowing blast-radius analysis of specific Maven packages across pull requests, default branches, and developer machines.

    3 details in Everything