Repository navigation
v3.11.0: who checks whose piece, verdict chips, the run's conversation - #106
Merged
Merged
Conversation
…tion - piece=<slug> in the [agt] header: a piece's reviewers and adversary hang under the executor that built it, subagent or pane - a finished reviewer row reads its verdict in its tone (PASS, FAIL P0:1, CONCERNS P1:2, APPROVE, REVISE, HELD, BROKEN 2) instead of 'done' - the deck's wire section is now the run's conversation: every dispatch (lead → role:piece, its task line) and every answer (its verdict or first line); a worker logs its real summary, not 'reported to the lead' - the band header carries the run's loops (plan ↺n · fix ×n), not the counts the status line already holds; in a run the spinner names the phases at work instead of repeating the count Refs #105 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A piece's executor, adversary and reviewers share piece=<slug>, which is all the mod needs to draw each reviewer under the build it checks. Optional: planning roles and single-piece runs leave it out. Refs #105 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
The deck pane is short; the conversation is what a person opens it for, so it now sits under the cast and the routing log follows. Refs #105 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
In the evals /agt never wrote piece=, and most runs have one executor. With exactly one builder on screen, a reviewer or adversary that names no piece now hangs under it. Two builders without pieces stay unlinked: the band does not guess which diff a reviewer reads. A plain subagent spawned after a run reads as a count on the spinner again, not 'agents'. Refs #105 Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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.
Closes #105
What
piece=<slug>in the dispatch header: a piece's reviewers and adversary hang under the executor that built it. A run with exactly one executor nests its reviewers without the field./agt-deckshows the run's conversation under the cast: every dispatch and every answer. A worker pane logs its real summary.plan ↺n · fix ×n) instead of the counts the status line holds. In a run, the spinner names the phases at work.Verification
claude plugin test .→ 371 pass, 0 fail ·bash scripts/validate.sh→ 0 fails ·claude plugin validate . --strict→ passedpiece=, including the two-slice case. That is why the one-executor fallback exists. Whether real multi-slice runs emitpiece=is still unverified.--plugin-dirsession with hand-written headers rather than through/agt: the code-reviewer of piece api was drawn under the api executor, its row turned intoFAIL P0:1, the spinner read· review, and/agt-deckshowed the four-line conversation (two tasks out, two answers back).🤖 Generated with Claude Code