--- title: "ci-speedup: Find & Fix Slow GitHub Actions | StarSling" description: "Free, open-source agent skill that finds what slows your GitHub Actions, pinpoints the check gating PRs, and gives your coding agent the evidence to fix it." url: https://starsling.dev/skills/ci-speedup canonicalUrl: https://starsling.dev/skills/ci-speedup --- # Fix slow GitHub Actions with /ci-speedup Install the agent skill to diagnose what's slowing down CI on your PRs and fix it. One required check often sets how long every PR waits to merge. ci-speedup samples your real Actions runs, finds that blocking check, and gives your coding agent the evidence to shorten it safely. Open source, MIT licensed, and runs on your machine. ## Install ```sh npx skills add starslingdev/skills ``` Then type `/ci-speedup` in your repository. ## What ci-speedup does **Deterministic core:** Detection, ranking, critical-path analysis, and every measured number are reproducible. **LLM boundary:** Only unmatched blocking logs get a grounded LLM read. Your coding agent uses the evidence and repository context to draft the fix. ### 01. Reads your workflows and code It opens your GitHub Actions workflow files and the code they build and test, so it understands what your CI is supposed to do and where the time should be going. ### 02. Analyzes your real runs and logs Through your own GitHub CLI it pulls your recent Actions runs, job timings, and logs. Real numbers from your repo, on your machine. Nothing is sent to StarSling. ### 03. Finds the bottleneck It works out which checks actually block a PR from merging and which one sets the total wait, then digs into that check step by step for the root cause. ### 04. Prompts your agent to fix it Every finding ends in a ready-to-paste prompt carrying the root cause, the log evidence, and the measured cost. Your coding agent checks the code and its git history first, so the fix is safe and intended, then writes the change for you to review. ## Example result In the example shown on the landing page, ci-speedup identifies E2E Tests as the blocking path and prompts the coding agent to shard it across four parallel jobs. Projected PR wait drops from 8m 24s to 3m 38s. ## Pattern catalog ci-speedup runs registered deterministic detectors drawn from a 78-pattern optimization catalog and elevates only causes supported by the measured runs. - [OPT2: Uncached large downloads](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt2--uncached-large-downloads) - [OPT4: Missing Docker layer cache](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt4--docker-layer-cache-missing) - [OPT12: Duplicated setup across jobs](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt12--duplicated-setup-across-jobs) - [OPT17: Sleep-based container readiness](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt17--sleep-based-container-readiness) - [OPT24: Long test job without sharding](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt24--long-test-job-without-sharding) - [OPT26: Outdated action major versions](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt26--outdated-action-major-versions) - [OPT28: Full git history checkout](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt28--full-git-history-checkout) - [OPT33: No draft PR gating on expensive jobs](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md#opt33--no-draft-pr-gating-on-expensive-jobs) [Browse all 78 patterns](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/optimization-patterns.md) ## Real fixes shipped in production ### ~3.4 minutes saved per run ci: shard E2E kitchen-sink across 3 parallel jobs Kitchen-sink E2E benchmark: 467s -> about 260s for the slowest shard across five warm-cache runs. StarSling split Mastra's E2E kitchen-sink workflow across a shard matrix and passed the shard number into Playwright. - [View PR #15888](https://github.com/mastra-ai/mastra/pull/15888) ### Customer case study [How Mastra made their slowest test suite 6x faster and cut queue time by 8x with StarSling](https://starsling.dev/customers/mastra) Their slowest test suite runs in 5 minutes instead of 30, and under load the job queue fell from about fifteen minutes to under two minutes. Every fix scoped and shipped by the agents. ### Local build caching restored chore: fix turbo cache configuration in ci Enabled local and remote Turborepo caching across three CI workflows. StarSling found that setting TURBO_CACHE to remote:rw disabled Turborepo's local cache, so .turbo was never written and the actions/cache save step had nothing to store; removing the override restored the local-plus-remote default so the build cache persists across runs. - [View PR #7950](https://github.com/better-auth/better-auth/pull/7950) ### Customer case study [How Better Auth got 2x faster E2E tests and cut 20,000 CI minutes a month with StarSling](https://starsling.dev/customers/better-auth) Per-job E2E from 2m 22s down to 1m 04s. Agent-native CI for the auth library every team's PR queue gates on. ## FAQ ### What is ci-speedup? ci-speedup is a free, open-source agent skill that analyzes your GitHub Actions workflows and recent runs, finds the merge-gating checks that set PR wait time, ranks their long poles, and produces an evidence-backed prompt for your coding agent to draft the targeted fix. ### Is ci-speedup free? Yes. ci-speedup is free, open source, and MIT licensed. You can install it from the public StarSling skills repository and run it in your own repository. ### Does ci-speedup send my code to StarSling? No. ci-speedup runs locally through your coding agent and uses your authenticated `gh` CLI to read GitHub Actions runs, timings, and logs. It does not send your code or CI data to StarSling. ### How does ci-speedup find what's slowing my GitHub Actions down? It deterministically identifies the required checks on the blocking path, measures recent job and step timings, and runs registered detectors from a 78-pattern CI optimization catalog. It explains evidence-backed causes and reports detector gaps honestly; your coding agent uses the repository context to draft the fix. ### Which coding agents does ci-speedup work with? ci-speedup works with all [73 coding agents currently supported by Vercel's Skills CLI](https://github.com/vercel-labs/skills#supported-agents), including OpenCode, Claude Code, Codex, and Cursor. ## Resources - [View the repository](https://github.com/starslingdev/skills) - [Read the skill](https://github.com/starslingdev/skills/tree/main/skills/ci-speedup) - [Read the measurement methodology](https://github.com/starslingdev/skills/blob/main/skills/ci-speedup/references/wall-clock-methodology.md)