Repository navigation
Report review execution state truthfully in user-facing output - #31
Merged
Merged
Conversation
Clean check summaries now distinguish a completed model review from runs where no model was called (review disabled by config, empty diff). Error envelopes carry observed wall-clock duration instead of a hardcoded zero, and failed model runs retain accumulated token usage across repair, cascade, and consensus paths instead of reporting zero tokens. Docs now match what the implementation measures: doctor's forge-token check is labeled presence-only, the README points at the published bench results instead of unbenchmarked per-model recommendations, and the bench docs name their metrics seeded-region hit rate and catalog-priced cost estimates, dropping unsupported severity-calibration and run-variance claims.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Clean check summaries now distinguish a completed model review from runs where no model was called (review disabled by config, or an empty diff), error envelopes carry observed wall-clock duration instead of a hardcoded zero, and failed model runs retain accumulated token usage across repair, cascade, and consensus paths. Doctor output labels its forge-token check as presence-only, the README points at the published bench results instead of unbenchmarked per-model recommendations, and the bench docs name their metrics for what they measure (seeded-region hit rate, catalog-priced cost estimates) and drop unsupported calibration and variance claims.