Skip to content

playbook 2.2.2

Choose a tag to compare

@andresfrei andresfrei released this 17 Nov 22:08
· 218 commits to main since this release
753e158

Version 2.2.2 makes the report of review-change and the questions of prepare-project closer to what each skill says, and makes their evaluation cases read the meaning of an answer. It breaks nothing.

What's fixed

  • review-change reports in a fixed order. After the change it reviewed, it says what the change is meant to do, then its findings. When there are none, it says "No findings." and does not list the cases that passed. Its description no longer asks the session to show the report unchanged, since Claude Code has no way to do that.
  • prepare-project no longer gives, as its example of a closing line that asks two things, the very question a new booking app calls for, which the agent then asked.
  • Evaluation cases accept other wordings of the same answer, in English and Spanish, and keep-minimal-triggers now reviews a commit in a repository.

Measured with claude plugin eval, three runs per case. What still fails is listed as open questions in the wiki, for a later version.

Update

claude plugin marketplace update playbook
claude plugin update playbook@playbook

Then update each of playbook-wiki@playbook and playbook-agents@playbook that claude plugin list shows. The changes work from the next session.

Licence

MIT. Issues and pull requests are welcome; see CONTRIBUTING.md.