Repository navigation
Grade what triage cannot see and add /codetrial-odds - #447
Merged
Merged
Conversation
A skipped gates lane passed as green, an approval survived the push after it, and resolving an outdated thread raised a grade nobody had earned, as #430 showed. The sweep now grades those, names everyone a PR waits on, and lists stacked prerequisites and PR pairs that change the same base lines. The review of a low or blocked PR rechecks its outdated threads, and its findings go out as one review, so the author gets one notification instead of one per finding.
Asking which open PRs could merge meant running the whole triage sweep. /codetrial-odds runs only that question: it refreshes a snapshot kept in the git directory, replaced only by a whole sweep, and reports each PR's grade, who it waits on, the merge order the sweep records and what moved since the last run. It is typed-only, so the plain-language request still routes to codetrial-triage.
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
A name holding " b/" split at the wrong place in a diff header, a renamed section took its path from that split, and the odds snapshot was one file for every repository, though the sweep matches PRs by number alone. The header now settles the split by its matching halves and a rename by its "rename to" line, the snapshot is kept per repository, the sweep resolves only threads the brief listed, and the odds report no longer reads a shared issue as a merge order.
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
The "rename to" line is the only thing that names a renamed file whose path holds " b/", since the diff --git line splits it at the wrong place. A test now fails when the sweep stops reading that line.
ctutils 0.4.3 generates masks and compares u128 values in non-constant time, and cargo-audit now fails every branch on it, main included. 0.4.5 is the fixed release; only the lockfile moves.
ColtenOuO
approved these changes
Oct 23, 2026
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 triage sweep graded only what it could see, and three things slipped past it: a skipped gates lane counted as green, an approval survived the push after it, and resolving an outdated thread raised a grade nobody had earned (#430 went from low to medium-high that way). The sweep now grades those, lists everyone a PR waits on, and records the merge order it can prove: stacked or stated prerequisites, and PR pairs whose diffs change the same base lines. The review of a low or blocked PR rechecks its outdated threads before any is resolved, and its findings go out as one review instead of one comment each. /codetrial-odds is a typed-only shortcut that reports merge confidence from a snapshot kept in the git directory, so the plain-language request still routes to codetrial-triage.
It also moves ctutils from 0.4.3 to 0.4.5 in Cargo.lock: RUSTSEC-2026-0385 (non-constant-time masks and u128 comparisons) was published while this was open and fails cargo-audit on main as well.
Verified with
scripts/test.sh portable(exit 0; 94 skill-script tests, 10 of them new), shellcheck,scripts/check-skill-refs.py, and live sweeps of this repository: 22 open PRs, about 45 s, 21 collisions that match the clashes found by hand (#411 and #442 inweb/devices.js, #422 and #430 inweb/compiler-explorer.js).GitHub accepts a COMMENT review with an empty body when it carries inline comments: probed once on this PR, then the probe comment was deleted.
post-inline.shstill falls back to one comment per finding when a review is refused.scripts/check-skill-routing.pyneeds a logged-in claude and was not run; no routed description changed.