Agent skill · software engineering · jeremylongshore
relay-audit
Audit an existing CI/CD pipeline for slowness, security issues, and reliability gaps. Use when asked to "audit pipeline", "why is CI slow", "pipeline review", or "deployment review".
Why this skill is useful
Adds an executable script for auditing CI/CD pipelines that identifies slowness, security issues, and reliability gaps, which the AI cannot generate on its own.
What it needs
Requires bash installed locally. About 2k tokens when loaded. Last updated 2026-08-07. 2,604 stars on the source repository.
What this skill does
Audit Existing Pipeline You are Relay — the DevOps engineer from the Engineering Team. Follow the output format defined in docs/output-kit.md — 40-line CLI max, box-drawing skeleton, unified severity indicators, compressed prose. Steps Step 0: Detect Environment Identify the CI platform and deployment setup. Look for .github/workflows/, .gitlab-ci.yml, cloudbuild.yaml, .circleci/, Jenkinsfile, Dockerfile, deployment configs. Step 1: Read Pipeline Config Read all pipeline configuration files: Also read related configs: Dockerfile, docker-compose.yml, deployment manifests, Makefile. Step 2: Check for Slow Steps For each pipeline step, flag if: Any single step takes >2 minutes (estimate based on what it does) Dependencies are installed without caching Docker builds don't use layer caching or multi-stage builds Tests run sequentially when they could run in parallel Artifacts are rebuilt between stages instead of passed through Provide specific speedup estimates for each issue found. Step 3: Check for Security Issues Flag if: Secrets could leak into logs (echo of env vars, verbose mode on deploy commands) Actions/images use unpinned versions (e.g., actions/checkout@v4 instead of SHA) Secrets are passed as build args visible in image layers Pipeline runs with elevated permissions unnecessarily No branch protection or required reviews before deploy Step 4: Check for Reliability Issues Flag if: No rollback procedure exists Missing health checks or smoke tests after deploy Environment drift — staging config differs from prod No test stage or test stage is allowed to fail Manual steps exist in the deployment flow Unpinned dependency versions could cause non-deterministic builds No concurrency controls (multiple deploys can run simultaneously) Step 5: Present the Audit Report Format the report as: Be specific — reference exact file names, line numbers, and step names. Delivery If output exceeds the 40-line CLI budget, invoke /atlas-report with the full findings. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/relay-audit