Agent skill · software engineering · jeremylongshore
relay-recon
Map the full CI/CD pipeline — triggers, build, test, deploy flow — with risk assessment. Use when asked "how does this deploy", "map the pipeline", or "understand CI/CD".
Why this skill is useful
Adds a comprehensive Bash script for mapping CI/CD pipelines with risk assessment that includes specific commands and documentation formats not commonly known.
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
Pipeline Reconnaissance 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, deployment targets, container configs, and infrastructure-as-code files. Step 1: Read All Pipeline Configs Read every pipeline and deployment configuration in the project: Also check for deployment configs: Kubernetes manifests, fly.toml, render.yaml, vercel.json, netlify.toml, app.yaml, terraform files. Step 2: Map the Pipeline Flow Trace the full path from code commit to production: 1. Trigger — what events start the pipeline (push, PR, tag, manual, schedule) 2. Build — how the artifact is produced (Docker build, npm build, go build, etc.) 3. Test — what tests run and what can fail silently 4. Deploy — how and where the artifact is deployed 5. Verify — any post-deploy checks (smoke tests, health checks) Step 3: Identify Key Details Document: Secrets locations — where secrets are referenced and what they're used for Deployment targets — all environments (dev, staging, prod) and their URLs/identifiers Manual steps — anything that requires human intervention Rollback capability — whether rollback exists and how to trigger it Average deploy time — estimate based on pipeline steps Branch strategy — what branches trigger what environments Step 4: Assess Risks Evaluate: Single points of failure in the pipeline Steps with no error handling or retry logic Missing stages (no tests, no smoke tests, no rollback) Blast radius of a bad deploy (all traffic at once vs. gradual) Recovery time estimate if something goes wrong Step 5: Present the Recon Report Format as: Factual and actionable. Map for someone taking over the project. Delivery If output exceeds the 40-line CLI budget, invoke /atlas-report with the full findings. The HTML report is the output. …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills jeremylongshore/relay-recon