docs(improve): enforce one-run-at-a-time in the swarm-mode cadence - #129
Merged
Conversation
Removed the "early rerun to verify" coordinator shortcut from the swarm-mode cadence, which could start a new exercise run while the swarm was still implementing the current batch's fixes -- wasting the run and making grading ambiguous about which fixes it was actually exercising. The coordinator now lands all entangled/seam-uncertain fixes first, dispatches the independent findings to the swarm, waits for the batch to fully complete, then does a single batch acceptance rerun. Includes the packaged mirror at plugins/autocoder/skills/improve/SKILL.md (tests/test_skill_packaging.sh requires the two stay identical). The other three fixes originally bundled in PR #113 (test_issue_source_lib.sh explicit-export syntax, worker-launch-lib.sh Gemini interactive mode, test_issues_gh_search.sh BASH_ENV clearing) are already present on current master -- each conflicted on cherry-pick with only a comment-wording difference, no functional diff. Reapplying them would have been a no-op.
4 tasks
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Split from #113, which bundled four unrelated changes:
Checked each independently on cherry-pick onto current master: (1), (2), and (4) already conflicted, and in every case the conflict was a comment-wording difference only -- the functional fix was already present on master (landed via other paths). Reapplying them would have been a no-op diff. Only (3) is a genuine, unlanded change, so it's the only one in this PR.
Change
Removes the "early rerun to verify" coordinator shortcut from the swarm-mode cadence, which could start a new exercise run while the swarm was still implementing the current batch's fixes -- wasting the run and making grading ambiguous. The coordinator now lands entangled/seam-uncertain fixes first, dispatches independent findings to the swarm, waits for the batch to fully complete, then does one batch acceptance rerun.
Verification
Confirmed (1), (2), (4) are no-ops: