Pipeline Doctor

Pipeline Doctor

Find what’s broken, unsafe, slow and expensive in your GitHub Actions workflow. Paste it below or drop the file.

0 lines

.github/workflows/*.yml

Paste a workflow, drop the file, or load the example

Loading editor…

Your workflow stays in your browser during Quick Scan — analysis runs locally, nothing is uploaded.

What Quick Scan checks

Every finding comes from one of 30 deterministic rules — a detector, an explanation and a recommended change. No guessing, no cloud round-trip: the whole rule set runs in this tab.

Security

14 rules

Performance

6 rules

Cost

4 rules

Reliability

4 rules

Maintainability

2 rules

Try it on a real scan

Loads a workflow assembled from the problems we keep finding in real repositories — over-scoped permissions, a pwn-able runner setup, dead matrix spend — and runs the same 30 rules your workflow will get.

Where this is going

Everything on this page runs locally today. The rest is built in the open — listed here because it is not in the product yet, and won’t be until it works.

  1. Phase 1

    Client-side Quick Scan — paste a workflow, get deterministic findings and verified fixesavailable

  2. Phase 2

    Go analyzer service with actionlint and zizmor behind a queue, sandboxed per job

  3. Phase 3

    GitHub App — connect a repository, discover workflows, deep scans

  4. Phase 4

    Run and job telemetry — real durations, flaky jobs, critical path with data

  5. Phase 5

    Cost analytics — runner minutes, waste detection, monthly estimates

  6. Phase 6

    Pull request checks — regressions before merge, with before/after diffs