Skip to content

Release 0.3.0 - #67

Merged
jt14den merged 1 commit into
mainfrom
release-0.3.0
Sep 30, 2026
Merged

jt14den merged 1 commit into
mainfrom
release-0.3.0

Conversation

@jt14den

@jt14den jt14den commented Sep 30, 2026

Copy link
Copy Markdown
Contributor

Bumps the version to 0.3.0 (checker/__init__.py, pixi.toml) and dates the CHANGELOG entry for #57, the standards review.

It's a minor version rather than a patch because the results file format moved to v3, which 0.2.1 can't read, and there are new explain and rules commands.

It also adds design/2026-09-30-ai-prompt-eval.md, the live check CONTRIBUTING asks for when the AI prompt changes. It compares 0.2.1 with the new prompt on the Claude backend: two real episodes, two runs each. Volume and reliability are the same. Glossary and accessibility findings are better hedged, and audience findings are phrased as assumed prerequisites. Accuracy findings are unchanged. The CHANGELOG now lists the local Ollama truncation bug (#66) as a known issue, and links the deferred items to #58-#63.

@jt14den
jt14den merged commit fdf97c1 into main Sep 30, 2026
3 checks passed
@jt14den
jt14den deleted the release-0.3.0 branch September 30, 2026 23:19
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant