farrellh1/evalpr

Eval-graded AI code review for your PRs

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
api_keyOpenRouter API keyyes
reviewer_modelOpenRouter model id for the revieweranthropic/claude-sonnet-4.6
grader_modelOpenRouter model id for the graderanthropic/claude-haiku-4.5
confidence_thresholdMinimum final_score (0-100) to retain a finding70
ignore_pathsComma-separated glob patterns to ignorenode_modules/**,*.lock,dist/**,build/**,*.min.js
max_filesSkip review if PR changes more files than this20
skip_draftsSkip draft PRstrue
config_pathPath to .evalpr.yml in the consumer repo.evalpr.yml
namedescription
retained_countNumber of high-confidence findings posted
hidden_countNumber of low-confidence findings hidden by threshold