Changelog

Version history

Every release of vouqis-verify. Unreleased changes are on GitHub.

v0.1.32026-07-01

doctor command added; baseline comparison now includes previous run score.

featvouqis doctor checks config validity, eval script path, and GitHub token before CI runs.
featBaseline score from the last passing run is included in the PR comment delta row.
fixvouqis verify now exits non-zero on BLOCK_MERGE verdict so GitHub Actions marks the step as failed.
choreRestore scroll position to top on every page load via inline head script.
v0.1.22026-06-18

PR comment posting via GitHub API; MERGE_WITH_WARNING verdict introduced.

featPost formatted verdict comment to pull request via GitHub REST API on every verify run.
featMERGE_WITH_WARNING verdict fires when score passes threshold but regresses from baseline.
featreport.block_on_fail config key controls whether a failing check blocks the merge button.
fixScore key lookup now raises a clear error when score_key is missing from eval output JSON.
v0.1.12026-06-05

Config validation hardening; ai_paths glob support.

featai_paths entries now support glob patterns (e.g. src/agents/**/*.py).
fixvouqis init validates that the eval script path exists before writing vouqis.yml.
refactorConfig loading extracted into its own module with typed schema validation.
chorepytest added to CI; config loader and verdict logic covered.
v0.1.02026-05-22

First public release. AI path detection, eval runner, three-verdict engine.

featvouqis verify detects changed files against ai_paths and skips runs on non-AI PRs.
featEval runner executes any shell command and reads score from JSON stdout.
featBLOCK_MERGE and SAFE_TO_MERGE verdicts based on configurable threshold.
featvouqis.yml config with ai_paths, eval.run, eval.score_key, eval.threshold.
featvouqis init interactive setup generates vouqis.yml from prompts.