Skip to content

cortex: accept numbered partial review files for a rolling slot (#341) - #7

Merged
Jammy2211 merged 1 commit into
mainfrom
feature/batch-status-box
Sep 3, 2026
Merged

Jammy2211 merged 1 commit into
mainfrom
feature/batch-status-box

Conversation

@Jammy2211

Copy link
Copy Markdown
Contributor

Summary

The Cortex board is rolling, but its review file was not: check demanded one batches/reviews/<slot>.md per record, so the first sitting closed the slot and a later arrival could only be ruled on by editing a ledger file that is already someone's evidence.

  • batch_problems resolves reviews/<slot>-r<N>.md (N >= 2) to the record batches/<slot>.md for the member check, accepts either the file's own stem or the slot's in the title, and refuses -r0/-r1 (the first review is <slot>.md).
  • A record's - review: key repeats, one line per review file, and every line must resolve; the old check saw only the first.
  • rule --batch <slot> needed no change: it validates the record, never the review file.
  • Doctrine: batches/AGENTS.md (a slot closes when nothing is left in a board state, not when the first review lands), batches/reviews/AGENTS.md (numbered-file grammar and title rule), batches/packets/TEMPLATE.md (submit URL commits to the next free review path), REFERENCE.md, schema decision 57.

Part of PyAutoLabs/PyAutoBrain#341 (batch phase 6: batch status box on both dashboards + rolling science review). Merge order: this PR first, then PyAutoBrain, then PyAutoMind — the Brain's numbered-review collect path must not land before cortex.py check accepts it.

Test plan

  • python -m pytest tests/ in PyAutoCortex
  • python scripts/cortex.py check clean on the worktree

🤖 Generated with Claude Code

https://claude.ai/code/session_01SgtS7NhqeziqgxKfSiMzRL

The Cortex board is rolling, but its review file was not: `check` demanded
one `batches/reviews/<slot>.md` per record, so the first sitting closed the
slot and a human who ruled on the members that had landed could only rule on
a later arrival by editing a ledger file that is already someone's evidence.

`batch_problems` now resolves `reviews/<slot>-r<N>.md` (N >= 2) to the record
`batches/<slot>.md` for the member check, accepts either the file's own stem
or the slot's in the title, and refuses `-r0`/`-r1` — the first review is
`<slot>.md`. A record's `- review:` key repeats, one line per review file,
and every line must resolve (`_record_values`); the old check saw only the
first. `rule --batch <slot>` needed no change: it validates the record, never
the review file, so a ruling traced to a `-r2` sitting already names the slot.

Doctrine: `batches/AGENTS.md` (a slot closes when nothing is left in a board
state, not when the first review lands), `batches/reviews/AGENTS.md` (the
numbered-file grammar and the title rule), `batches/packets/TEMPLATE.md` (the
submit URL commits to the next free review path), `REFERENCE.md` and schema
decision 57.

Cites PyAutoBrain#341.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01VsKeuX83FNGjLqs2axLHeR
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.

1 participant