Repository navigation
Fail closed on narrated risk; render the product statusline - #11
Merged
Merged
Conversation
A summary that describes merge-relevant risk alongside an empty findings array previously produced a clean, silent review - the predecessor product posted exactly this contradiction on 9 of its last 16 'clean' reviews. The contract now requires summary and findings to agree: the prompt states it, the client gives the model one corrective retry to either structure the risk or retract it, and a surviving contradiction fails closed with a synthetic finding that preserves the narrated text. The envelope additionally records the ungrounded-finding count and the engine wall-clock duration for quality telemetry.
PR comments and check summaries now carry the brand status icons (postil.dev/status/*.svg) with the 'confidence 0.91 - kind: risk' metadata line the product page shows, and confidence renders as a bare decimal instead of a percentage. Terminal output gains the matching glyphs, per-finding confidence/kind trailer, a findings/suppressed tally, and the promised lowercase 'gate: failing (fail-on: error)' line.
|
Postil could not validate the model output. Postil status: needs-attention | errors=1 warnings=0 info=0 inline_comments=1 |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The model narrating merge-relevant risk in its summary while returning zero structured findings now fails the review closed (after one corrective retry) instead of passing as a clean, silent run - the predecessor bot posted exactly that contradiction on 9 of its last 16 clean reviews. PR comments, check summaries, and TTY output now match the statusline the product page promises (brand status icons, decimal confidence, kind metadata, lowercase gate line), and the envelope records ungrounded-finding counts plus engine duration for quality telemetry. Covered by two new e2e cases; the clean-diff fixtures were corrected to honor the summary/findings consistency contract.