The GitHub Actions knowledge system
Make GitHub Actions fast, cheap, and safe
34 public rules, each with a page that says how to detect the problem, how to fix it, and how to verify the fix: readable by you, fetchable by your coding agent, and applied automatically by StarSling.
How StarSling worksStart here
- The rule catalog
GitHub Actions optimizations
Every CI rule StarSling checks: performance, reliability, cost, and security, each with detection steps, the fix, and an agent prompt.
- Diagnosis
Diagnose a GitHub Actions problem
Start from the symptom you have: slow tests, queued runs, cache misses, stuck jobs. Each page names the likely causes in order and links the fix.
- The practice guides
GitHub Actions best practices
Eighteen reference guides on what good CI looks like: caching, sharding, trigger scope, runners, and workflow security, with real-world exemplars.
- Workflow patterns
docker/build-push-action, copy-paste ready
A complete Docker build-and-push workflow with layer caching, explained line by line.
- Workflow patterns
Docker CI on GitHub Actions, end to end
Building, testing, and shipping images from CI: the whole pipeline in one page.
For coding agents
These rules ship as open-source agent skills that carry the detectors rather than describing them: ci-speedup, ci-secure and ci-score, installable from the skills page.
The whole catalog is one fetch: /github-actions/catalog.json sets out both ways to use it: run the skills, or apply the rules straight from these pages with nothing installed. It asks your agent which one you want before it installs anything, and the command that installs the skills lives inside the option that installs. Then it names the three skills and lists every rule with its stable ci.* id, category, detection mode, and page URLs. Every page on this site also serves a plain .md version for agents, listed in /llms.txt.
Last updated 2026-08-31