Works with the AI Agents you already use
CI/CD Security Audit — GitHub Actions
This skill finds all of them in one pass and hands back a triaged, file:line fix list.
$19
About this skill
The same handful of workflow bugs accounts for most GitHub Actions compromises — a tag-pinned action gets re-pointed at malicious code, a PR title lands in a run: block, pull_request_target checks out the head SHA with secrets in scope. This skill finds all of them in one pass and hands back a triaged, file:line fix list. The agent runs a bundled, zero-dependency Node script over .github/workflows/*.yml and turns the JSON report into a hardening plan. DETECTS (fail severity) - script-injection — attacker-controlled github.event.* fields (PR/issue titles and bodies, comments, reviews, head_ref, commit messages) interpolated into run blocks - pull-request-target-checkout — ref: github.event.pull_request.head.sha under pull_request_target — the textbook CI RCE - permissions-write-all — every scope granted to GITHUB_TOKEN - unpinned-action — uses: with no version ref at all DETECTS (warn severity) - unpinned-action — uses: pinned to a mutable tag/branch instead of a full commit SHA; docker:// images without a @sha256: digest - pull_request_target trigger present - permissions-missing — no permissions: block declared - permissions-broad-write — contents: write / id-token: write - secret-in-run — ${{ secrets.* }} interpolated directly in run/echo lines OUTPUT JSON on stdout: findings[] with id, severity, file, line, detail, recommendation, plus a summary with fail/warn counts and byId breakdown. Exit 0 clean / 1 findings / 2 usage error — drop into a pre-merge check or CI job. WHY IT'S SAFE TO RUN EVERYWHERE Pure line-based static scan: reads *.yml/*.yaml only, zero npm dependencies, no network calls, never executes a line of workflow code. Node 18+. HONEST ABOUT LIMITS A line-based auditor can't evaluate action internals, reusable-workflow trust, environment protection rules, or org/enterprise token defaults. The skill's triage guide says exactly what was and wasn't checked — clean means "no known patterns matched", not "secure". Includes a triage playbook per finding ID and an optional CI-gate recipe (fail the pipeline on summary.fail > 0).
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 21 days ago