Repository navigation
chat: settings says checker, no zero durations or byte counts, and help and manual match the screen - #1711
Merged
Conversation
Settings exposed the internal checking role identifier on its row and model picker, and searching by the crew seat word did not find it. The row, search and picker now use checker from the crew seat vocabulary while persisted role pins and model resolution retain their identifiers. Regression tests cover the rendered row, seat-word search and picker identity. The roles manual is updated in the final corpus commit of this wave because commands.md also carries the help and tab corrections. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Settled thinking rows printed zero seconds and could derive a duration from an absent timestamp. Rows now omit a duration until at least one second has been measured, retain positive token estimates and the disclosure, and keep the existing rounded display for longer spans. The duration-free sentence keeps the collapsed thought for wording required by the existing design tests. Deterministic timestamp cases and the manual describe that sentence and its zero-token form. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
The plain-folder snapshot recorder deliberately supplies no patch text, but senior-dev described that missing measurement as zero bytes. Candidate descriptions now show the changed files and tree without a byte size when no positive size exists, and continue to print measured positive sizes. Regression tests use the real snapshot recorder and retain a positive-size control. The manual explains why a plain-folder ending has no byte count. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
The combined patch and doc synopsis left doc without a leading command line, so its own help fell back to the bare command name. Per-command help now reads each explicitly named command on a shared row separately, and doc help prints codeaf doc PATH [--pages A-B]. The front page keeps its compact shared row and existing line cap, and pipes in arguments or alternate forms remain intact. The regression test reaches runDoc's help door, and the existing eighty-column law covers all help pages. The shell manual explicitly describes the corrected help output. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
…reen The manual still promised the environment table in the main help page, omitted Teams from both tab lists, promised Home on every keyed launch and invented an untitled-manager handle. Each retrieved section now describes the running screen, quotes the current help footer and explains the team-name or lead fallback. The roles section also calls the checking row checker to match the Settings correction in this wave. New corpus-wide stale-claim checks and positive section checks were appended without changing any existing test. CONTRACT.md records the observable behaviours and preserved design choices. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
…belongs to its submission receipt The role table under /settings Providers still listed the checking role as auditor, which the screen no longer shows; it now says checker and notes that a saved pin keeps the old id. The senior-dev page credited the across-N-files wording to the run's ending, but that wording is the submission receipt's; the last line is the ending sentence. The corpus guard now also catches the old table row. Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
Co-Authored-By: Claude Opus 5.5 (1M context) <[email protected]>
AbirAbbas
marked this pull request as ready for review
October 1, 2026 02:05
santoshkumarradha
added
the
area:chat
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.
What was wrong
These are small truth defects the re-verification pass over dev (
feac16a48) found. All of them were already in v0.5.0./settings→ Providers labelled the checking role with its internal id,auditor, a banned machinery word.thought for 0s · 10 tok.0 bytes across 3 file(s), although a plain folder produces no patch text to measure.codeaf doc --helpprintedcodeaf docwith noPATH.codeaf --helpprints the environment table (it points tocodeaf help env).Teams).@manager-…handle (it is@<team>, or@leadfor All teams).What is true now
checkereverywhere, the way/crewdoes. A saved pin keeps theauditorid.codeaf doc --helpreadscodeaf doc PATH [--pages A-B], and every help page stays within 80 columns.internal/manual/chatdisplaytruth_test.gokeeps each stale sentence from coming back.Proof
Each code item has a test that fails before the change (
settings_role_words_test.go,thinking_duration_test.go,solo_patch_size_test.go,doc_help_test.go), plus tmux frames for the Settings label and the thinking row. No existing test was changed. An independent read-only review found two remaining manual mismatches, and the last commit fixes them.🤖 Generated with Claude Code