CI Playbooks
Lessons from software factories
- Techniques
- 253
- Merged PRs
- 226
- Playbooks
- 10
- Last updated
Agent-friendly playbooks for GitHub Actions
Each playbook is based on an analysis of each company's public posts, talks, commits and pull requests, as available.
- Read Anthropic’s playbookRun only affected tests
- Read DoorDash’s playbookRun build and test in parallel, sharded across CI agents
- Read Linear’s playbookBenchmark the same job on faster hardware
- Read Mastra’s playbookStarSling runners for the slowest test workflow
- Read OpenAI’s playbookRank CI fixes by the minutes they save
- Read Ramp’s playbookMove a CPU-bound test suite to a faster CPU
- Read Shopify’s playbookLet an agent iterate against one CI metric
- Read Spotify’s playbookBuild arm64 on native arm64 runners
- Read Stripe’s playbookRun only the tests a change can reach
- Read Uber’s playbookBuild the merge-queue speculations most likely to be needed
Top techniques across playbooks
ci-speedup patterns that match techniques in more than one playbook.
- Cache dependenciesOPT61Used in 9 of 10 playbooks
- Measure test concurrencyOPT88Used in 9 of 10 playbooks
- Shard long test jobsOPT24Used in 8 of 10 playbooks
- Run independent steps in parallelOPT83Used in 7 of 10 playbooks
- Move slow jobs to faster runnersOPT81Used in 6 of 10 playbooks
- Remove duplicate commandsOPT16Used in 5 of 10 playbooks
- Balance test shardsOPT25Used in 5 of 10 playbooks
- Check out shallow historyOPT28Used in 5 of 10 playbooks
- Run only the affected appsOPT40Used in 5 of 10 playbooks
- Cancel superseded runsOPT46Used in 5 of 10 playbooks
- Set job timeoutsOPT57Used in 5 of 10 playbooks
- Cache large downloadsOPT2Used in 4 of 10 playbooks
From StarSling
Self-improving CI
StarSling runners make GitHub Actions faster from the first job. On paid plans, StarSling agents keep finding CI improvements like these and open them as pull requests for your team to review.