Skip to content

prompt: worktree-claim-parser-forms — issue, registration and completion record - #154

Merged
Jammy2211 merged 2 commits into
mainfrom
claude/automind-task-planning-ef00l2
Aug 8, 2026
Merged

prompt: worktree-claim-parser-forms — issue, registration and completion record#154
Jammy2211 merged 2 commits into
mainfrom
claude/automind-task-planning-ef00l2

Conversation

@Jammy2211

Copy link
Copy Markdown
Collaborator

Mind bookkeeping for PyAutoBrain#210 (squash-merged 83b65f1c) / PyAutoBrain#209 (closed completed).

The task fixed worktree_list_claimed, whose awk split active.md claim bullets on ": " only — so the paren form - <repo> (<branch>) produced a malformed repo name and worktree_check_conflict, the start_dev step-6 guard, returned "no conflict" for roughly half of all real claims (history runs 162 colon-form to 139 paren-form).

Contents

Two commits, the full draft/ → active/ → complete/ lifecycle for one task:

  1. Issue + registration — prompt advanced draft/bug/pyautobrain/active/, task registered in active.md.
  2. Completion recordcomplete/2026/08/worktree-claim-parser-forms.md written via lifecycle.py record with the prompt folded in, active.md section dropped, complete/index.md refreshed (935 records).

lifecycle.py check: OK. lifecycle.py index --check: OK.

Worth keeping from the record

  • Blast radius wider than the prompt stated. It scoped the bug to routing; repo_cleanup's CLAIMED protection set is also built from (repo, branch) pairs from this function, so in-flight branches were short one of their three delete-guards. IN_WORKTREE and OPEN_PR still covered them — nothing lost.
  • A shallow-clone trap. A first history sweep reported zero paren-form claims, appearing to refute the prompt outright. That was a cloud-session shallow clone showing two days, not the record. Check git rev-parse --is-shallow-repository before concluding anything from repo history.
  • Sizing not taken. Bug Agent scored large (8), prompt header said small; the header was right (41-line awk rewrite + tests). Same prose-driven inflation as the mge-sigma-min-workspace-sweep precedent.
  • Follow-up left open. Several active.md entries carry repos-none-claimed: notes written to steer around a guard that never fired — harmless now, no longer necessary, deliberately not swept here.

Registration departures (recorded, not silent)

This ran in a cloud session, so the entry recorded worktree: none and the session's designated branch rather than the feature/<task-name> + ~/Code/PyAutoLabs-wt/ convention start_dev prescribes.

🤖 Generated with Claude Code

https://claude.ai/code/session_016VnfCyAwgQdm1V3iJBtiga


Generated by Claude Code

claude added 2 commits August 8, 2026 13:54
Advance the prompt draft/ -> active/ and register the task in active.md.

The active.md claim parser fix is implemented and pushed on PyAutoBrain
claude/automind-task-planning-ef00l2; PyAutoBrain#209 tracks it.

The entry deliberately uses the paren claim form `  - PyAutoBrain (<branch>)` —
the form the guard could not previously see. With the fix in place
worktree_check_conflict now reports the conflict against this entry, which is
the first time the step-6 guard has fired on the real ledger.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016VnfCyAwgQdm1V3iJBtiga
PyAutoBrain#210 squash-merged as 83b65f1c; issue #209 auto-closed.

Records complete/2026/08/worktree-claim-parser-forms.md, folds in the
originating prompt, drops the active.md section and refreshes the index.

lifecycle.py check: OK. lifecycle.py index --check: OK.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016VnfCyAwgQdm1V3iJBtiga
@Jammy2211
Jammy2211 merged commit 8b33239 into main Aug 8, 2026
1 check passed
Jammy2211 pushed a commit that referenced this pull request Aug 21, 2026
…7, PR #458)

Exact 3.0-3.4x windowed numba kernel shipped for review; Rectangular euclid
3.08 -> 1.17 s / hst ~30 -> 10.1 s stacked with the merged wins. Default-CPU-mesh
strategic question intaken as autolens_profiling#153 + inference-programme
Phase 14 (ledger PR #154). Prompt advances draft -> active.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Vcc7MUBMnNU6n8qqS9ioVZ
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants