GitHub Actions security review
Audits GitHub Actions workflows for attacks an outsider without write access can run: pwn requests, expression injection, comment commands, token theft, poisoned agent configs, unpinned actions. Use when reviewing CI security or changing workflows.
gha-security-review- Author
- Sentry
- Source
- getsentry/skills/skills/gha-security-review @ a582c51
- License
- Apache-2.0 · Copyright 2025 Functional Software, Inc. dba Sentry
About this skill
What it does
The skill audits GitHub Actions workflows for holes an outside attacker can exploit, and reports only what it can show as an attack. The attacker has no write access: they can open fork PRs, file issues and comment, but cannot push or start workflow_dispatch. Anything that needs write access, such as dispatch inputs or push-only workflows on protected branches, is out of scope by design.
It reads .github/workflows/, composite and local actions, and files the workflows load: Makefile, scripts under .github/, CLAUDE.md, AGENTS.md. Eight checks follow: pwn requests (pull_request_target that checks out fork code), ${{ }} injection in run: blocks, comment commands without an author_association check, elevated tokens reachable by untrusted code, poisoned config files fed to AI agents, unpinned third-party actions in privileged jobs, permissions and secrets, self-hosted runners, caches and artifacts. A table of safe patterns, such as ${{ }} in if: or with: and first-party actions/* on tags, keeps false positives out.
A finding is HIGH when the agent has all five parts — entry point, payload, execution mechanism, impact, PoC sketch — and MEDIUM, under "Needs verification", when a link is missing. The report numbers findings GHA-001 onward, gives fix code and lists the workflows cleared.
Nine references are loaded by trigger: one per class (pwn-request.md, expression-injection.md, comment-triggered-commands.md, credential-escalation.md, ai-prompt-injection-via-ci.md, supply-chain.md, permissions-and-secrets.md, runner-infrastructure.md) and real-world-attacks.md, seven documented attacks from the HackerBot Claw campaign, used to calibrate severity.
When to use it, and when not
Use it on a repository that accepts outside PRs, before changing workflows, after adding an AI agent to CI, or as a one-off audit.
It ignores insiders with write access, workflows in other repositories and other CI systems, and it does not write or debug pipelines.
How to run it
The install command below puts the folder in place. The agent picks the skill up when asked about workflow or CI security; to call it yourself:
- Claude Code:
/gha-security-review, or ask for the skill by name. - Codex:
$gha-security-review. - Cursor: type
/in the Agent chat and pickgha-security-review.
Example: /gha-security-review Audit .github/workflows; we just added a Claude review job on pull_request_target.
How it differs from similar skills
GitHub Actions CI from our catalog writes and fixes workflows with SHA pins, minimal permissions and no pull_request_target for forks as defaults, but traces no attacks. Security review, Sentry covers application code and hands workflow files to this skill when it is installed; the two pair up. Security audit for codebases and services treats CI as one domain of a wider audit. Its supply-chain file covers untrusted code in privileged workflows, expression injection and cache or artifact trust for any CI system, with nothing specific to GitHub Actions.
Install
Run one command — it saves SKILL.md and the files it uses into your agent's skills folder. Claude Code and Codex pick up a new skill without a restart; if it does not show up, restart the agent.
curl -fsSL --create-dirs -o ~/.claude/skills/gha-security-review/SKILL.md https://anymodel.org/en/skills/sentry/gha-security-review/SKILL.md -o ~/.claude/skills/gha-security-review/LICENSE https://anymodel.org/en/skills/sentry/gha-security-review/LICENSE -o ~/.claude/skills/gha-security-review/references/ai-prompt-injection-via-ci.md https://anymodel.org/en/skills/sentry/gha-security-review/references/ai-prompt-injection-via-ci.md -o ~/.claude/skills/gha-security-review/references/comment-triggered-commands.md https://anymodel.org/en/skills/sentry/gha-security-review/references/comment-triggered-commands.md -o ~/.claude/skills/gha-security-review/references/credential-escalation.md https://anymodel.org/en/skills/sentry/gha-security-review/references/credential-escalation.md -o ~/.claude/skills/gha-security-review/references/expression-injection.md https://anymodel.org/en/skills/sentry/gha-security-review/references/expression-injection.md -o ~/.claude/skills/gha-security-review/references/permissions-and-secrets.md https://anymodel.org/en/skills/sentry/gha-security-review/references/permissions-and-secrets.md -o ~/.claude/skills/gha-security-review/references/pwn-request.md https://anymodel.org/en/skills/sentry/gha-security-review/references/pwn-request.md -o ~/.claude/skills/gha-security-review/references/real-world-attacks.md https://anymodel.org/en/skills/sentry/gha-security-review/references/real-world-attacks.md -o ~/.claude/skills/gha-security-review/references/runner-infrastructure.md https://anymodel.org/en/skills/sentry/gha-security-review/references/runner-infrastructure.md -o ~/.claude/skills/gha-security-review/references/supply-chain.md https://anymodel.org/en/skills/sentry/gha-security-review/references/supply-chain.md
curl -fsSL --create-dirs -o ~/.agents/skills/gha-security-review/SKILL.md https://anymodel.org/en/skills/sentry/gha-security-review/SKILL.md -o ~/.agents/skills/gha-security-review/LICENSE https://anymodel.org/en/skills/sentry/gha-security-review/LICENSE -o ~/.agents/skills/gha-security-review/references/ai-prompt-injection-via-ci.md https://anymodel.org/en/skills/sentry/gha-security-review/references/ai-prompt-injection-via-ci.md -o ~/.agents/skills/gha-security-review/references/comment-triggered-commands.md https://anymodel.org/en/skills/sentry/gha-security-review/references/comment-triggered-commands.md -o ~/.agents/skills/gha-security-review/references/credential-escalation.md https://anymodel.org/en/skills/sentry/gha-security-review/references/credential-escalation.md -o ~/.agents/skills/gha-security-review/references/expression-injection.md https://anymodel.org/en/skills/sentry/gha-security-review/references/expression-injection.md -o ~/.agents/skills/gha-security-review/references/permissions-and-secrets.md https://anymodel.org/en/skills/sentry/gha-security-review/references/permissions-and-secrets.md -o ~/.agents/skills/gha-security-review/references/pwn-request.md https://anymodel.org/en/skills/sentry/gha-security-review/references/pwn-request.md -o ~/.agents/skills/gha-security-review/references/real-world-attacks.md https://anymodel.org/en/skills/sentry/gha-security-review/references/real-world-attacks.md -o ~/.agents/skills/gha-security-review/references/runner-infrastructure.md https://anymodel.org/en/skills/sentry/gha-security-review/references/runner-infrastructure.md -o ~/.agents/skills/gha-security-review/references/supply-chain.md https://anymodel.org/en/skills/sentry/gha-security-review/references/supply-chain.md
foreach ($f in 'SKILL.md','LICENSE','references/ai-prompt-injection-via-ci.md','references/comment-triggered-commands.md','references/credential-escalation.md','references/expression-injection.md','references/permissions-and-secrets.md','references/pwn-request.md','references/real-world-attacks.md','references/runner-infrastructure.md','references/supply-chain.md') { $p = Join-Path "$HOME\.claude\skills\gha-security-review" $f; New-Item -ItemType Directory -Force (Split-Path $p) | Out-Null; Invoke-WebRequest "https://anymodel.org/en/skills/sentry/gha-security-review/$f" -UseBasicParsing -OutFile $p }foreach ($f in 'SKILL.md','LICENSE','references/ai-prompt-injection-via-ci.md','references/comment-triggered-commands.md','references/credential-escalation.md','references/expression-injection.md','references/permissions-and-secrets.md','references/pwn-request.md','references/real-world-attacks.md','references/runner-infrastructure.md','references/supply-chain.md') { $p = Join-Path "$HOME\.agents\skills\gha-security-review" $f; New-Item -ItemType Directory -Force (Split-Path $p) | Out-Null; Invoke-WebRequest "https://anymodel.org/en/skills/sentry/gha-security-review/$f" -UseBasicParsing -OutFile $p }Any other agent: save the file into its skills folder, or give it the link to the file: https://anymodel.org/en/skills/sentry/gha-security-review/SKILL.md
SKILL.md
<!-- Attack patterns and real-world examples sourced from the HackerBot Claw campaign analysis by StepSecurity (2025): https://www.stepsecurity.io/blog/hackerbot-claw-github-actions-exploitation -->
GitHub Actions Security Review
Find exploitable vulnerabilities in GitHub Actions workflows. Every finding MUST include a concrete exploitation scenario — if you can't build the attack, don't report it.
This skill encodes attack patterns from real GitHub Actions exploits — not generic CI/CD theory.
Scope
Review the workflows provided (file, diff, or repo). Research the codebase as needed to trace complete attack paths before reporting.
Files to Review
.github/workflows/*.yml— all workflow definitionsaction.yml/action.yaml— composite actions in the repo.github/actions/*/action.yml— local reusable actions- Config files loaded by workflows:
CLAUDE.md,AGENTS.md,Makefile, shell scripts under.github/
Out of Scope
- Workflows in other repositories (only note the dependency)
- GitHub App installation permissions (note if relevant)
Threat Model
Only report vulnerabilities exploitable by an external attacker — someone without write access to the repository. The attacker can open PRs from forks, create issues, and post comments. They cannot push to branches, trigger workflow_dispatch, or trigger manual workflows.
Do not flag vulnerabilities that require write access to exploit:
workflow_dispatchinput injection — requires write access to trigger- Expression injection in
push-only workflows on protected branches workflow_callinput injection where all callers are internal- Secrets in
workflow_dispatch/schedule-only workflows
Confidence
Report only HIGH and MEDIUM confidence findings. Do not report theoretical issues.
| Confidence | Criteria | Action |
|---|---|---|
| HIGH | Traced the full attack path, confirmed exploitable | Report with exploitation scenario and fix |
| MEDIUM | Attack path partially confirmed, uncertain link | Report as needs verification |
| LOW | Theoretical or mitigated elsewhere | Do not report |
For each HIGH finding, provide all five elements:
- Entry point — How does the attacker get in? (fork PR, issue comment, branch name, etc.)
- Payload — What does the attacker send? (actual code/YAML/input)
- Execution mechanism — How does the payload run? (expression expansion, checkout + script, etc.)
- Impact — What does the attacker gain? (token theft, code execution, repo write access)
- PoC sketch — Concrete steps an attacker would follow
If you cannot construct all five, report as MEDIUM (needs verification).
Step 1: Classify Triggers and Load References
For each workflow, identify triggers and load the appropriate reference:
| Trigger / Pattern | Load Reference |
|---|---|
pull_request_target |
references/pwn-request.md |
issue_comment with command parsing |
references/comment-triggered-commands.md |
${{ }} in run: blocks |
references/expression-injection.md |
| PATs / deploy keys / elevated credentials | references/credential-escalation.md |
| Checkout PR code + config file loading | references/ai-prompt-injection-via-ci.md |
| Third-party actions (especially unpinned) | references/supply-chain.md |
permissions: block or secrets usage |
references/permissions-and-secrets.md |
| Self-hosted runners, cache/artifact usage | references/runner-infrastructure.md |
| Any confirmed finding | references/real-world-attacks.md |
Load references selectively — only what's relevant to the triggers found.
Step 2: Check for Vulnerability Classes
Check 1: Pwn Request
Does the workflow use pull_request_target AND check out fork code?
- Look for
actions/checkoutwithref:pointing to PR head - Look for local actions (
./.github/actions/) that would come from the fork - Check if any
run:step executes code from the checked-out PR
Check 2: Expression Injection
Are ${{ }} expressions used inside run: blocks in externally-triggerable workflows?
- Map every
${{ }}expression in everyrun:step - Confirm the value is attacker-controlled (PR title, branch name, comment body — not numeric IDs, SHAs, or repository names)
- Confirm the expression is in a
run:block, notif:,with:, or job-levelenv:
Check 3: Unauthorized Command Execution
Does an issue_comment-triggered workflow execute commands without authorization?
- Is there an
author_associationcheck? - Can any GitHub user trigger the command?
- Does the command handler also use injectable expressions?
Check 4: Credential Escalation
Are elevated credentials (PATs, deploy keys) accessible to untrusted code?
- What's the blast radius of each secret?
- Could a compromised workflow steal long-lived tokens?
Check 5: Config File Poisoning
Does the workflow load configuration from PR-supplied files?
- AI agent instructions:
CLAUDE.md,AGENTS.md,.cursorrules - Build configuration:
Makefile, shell scripts
Check 6: Supply Chain
Are third-party actions securely pinned to full SHAs?
- Pin third-party / external actions and reusable workflows only
- Do not flag first-party
actions/*orgithub/*on version tags - Do not flag same-repo / vendored (
./.github/actions/...) as supply-chain pinning issues - Only report when the job has secrets, OIDC, write token, release, deploy, package, or signing power — unprivileged read-only CI is not a finding
Check 7: Permissions and Secrets
Are workflow permissions minimal? Are secrets properly scoped?
Check 8: Runner Infrastructure
Are self-hosted runners, caches, or artifacts used securely?
Safe Patterns (Do Not Flag)
Before reporting, check if the pattern is actually safe:
| Pattern | Why Safe |
|---|---|
pull_request_target WITHOUT checkout of fork code |
Never executes attacker code |
${{ github.event.pull_request.number }} in run: |
Numeric only — not injectable |
${{ github.repository }} / github.repository_owner |
Repo owner controls this |
${{ secrets.* }} |
Not an expression injection vector |
${{ }} in if: conditions |
Evaluated by Actions runtime, not shell |
${{ }} in with: inputs |
Passed as string parameters, not shell-evaluated |
| Third-party actions pinned to full SHA | Immutable reference |
First-party actions/* / github/* on version tags |
Outside third-party pinning policy — do not flag |
| Same-repo / vendored local actions | Not third-party supply chain (review pwn-request separately) |
pull_request trigger (not _target) |
Runs in fork context with read-only token |
Any expression in workflow_dispatch/schedule/push to protected branches |
Requires write access — outside threat model |
Key distinction: ${{ }} is dangerous in run: blocks (shell expansion) but safe in if:, with:, and env: at the job/step level (Actions runtime evaluation).
Step 3: Validate Before Reporting
Before including any finding, read the actual workflow YAML and trace the complete attack path:
- Read the full workflow — don't rely on grep output alone
- Trace the trigger — confirm the event and check
if:conditions that gate execution - Trace the expression/checkout — confirm it's in a
run:block or actually references fork code - Confirm attacker control — verify the value maps to something an external attacker sets
- Check existing mitigations — env var wrapping, author_association checks, restricted permissions, SHA pinning
If any link is broken, mark MEDIUM (needs verification) or drop the finding.
If no checks produced a finding, report zero findings. Do not invent issues.
Step 4: Report Findings
## GitHub Actions Security Review
### Findings
#### [GHA-001] [Title] (Severity: Critical/High/Medium)
- **Workflow**: `.github/workflows/release.yml:15`
- **Trigger**: `pull_request_target`
- **Confidence**: HIGH — confirmed through attack path tracing
- **Exploitation Scenario**:
1. [Step-by-step attack]
- **Impact**: [What attacker gains]
- **Fix**: [Code that fixes the issue]
### Needs Verification
[MEDIUM confidence items with explanation of what to verify]
### Reviewed and Cleared
[Workflows reviewed and confirmed safe]
If no findings: "No exploitable vulnerabilities identified. All workflows reviewed and cleared."
Connect through AnyModel API
This skill runs on whatever model your agent uses. Point Claude Code, Codex or Cursor at AnyModel to run it on Claude, GPT, Gemini and more with one key — pay per token, no subscription.
AnyModel