← Speckl

AI release readiness

Ten yes or no questions about how your team ships code an agent wrote. The score is calculated in this browser. Answers stay on this page.

The ten checks

0 of 10 answered

  1. 1 · Spec or acceptance criteriaBefore an agent starts on a change, does someone write a spec or acceptance criteria?
  2. 2 · Tests a human wrote or checkedDoes a human write the tests for the change, or read and check tests the agent wrote?
  3. 3 · Diff size limitsDo you limit how large an agent diff can get before it has to be split?
  4. 4 · Human review of agent pull requestsDoes a human review every pull request an agent opens before it merges?
  5. 5 · Secrets and dependency checksBefore release, do you check the diff for secrets and for new or changed dependencies?
  6. 6 · Staged rollout or feature flagDoes the change go out behind a feature flag, or to a small set of users first?
  7. 7 · Rehearsed rollbackHas someone rehearsed the rollback for this change?
  8. 8 · Monitoring after releaseAfter release, does someone watch errors, latency, or the metric this change should move?
  9. 9 · Agent instruction fileDoes the repo have an agent instruction file, such as AGENTS.md, that says how changes should be made?
  10. 10 · Written release checklistDo you keep a written release checklist and run it before an agent change goes to production?

AI Code Release Checklist

This page is the short version. The checklist is 18 checks on one page, for code written by people or by agents. Free, pay what you want. The checklist is free under CC BY 4.0. Share and adapt it, with credit.

The Agent Gate Pack is free as well.