docs(planning): staleness sweep disposition and wd.todo regroom

[djradon]

Stacked on #36 — merge that first and this reduces to the groom commit alone.

First sweep of the open-task back catalogue. A codex read-only pass classified all 41 open wa.task.* notes against main at 5cc3f40 by code evidence rather than checkboxes: 7 SHIPPED, 4 SUPERSEDED, 27 LIVE, 3 UNCLEAR, none skipped.

Archive-side renames landed separately at semantic-flow/weave-dev-archive@48d6657.

Why this was needed

The closure duty only ever fired on tasks the planning seat closed deliberately. Nothing had ever swept the back catalogue, and on 2026-08-01 an unrelated analysis stumbled on a July-shipped task still filed as open — by luck. This replaces the luck.

The conflicts that mattered

The sweep compared each note against wd.todo against the code, and reported disagreements:

• wd.todo still listed multi-pending first-payload weave as missing. It shipped in d6f87ca — the change that makes v0.7.0 a minor rather than a patch.
• Several lines read as though the append planner did not exist. planInventoryAppend has implemented no-op/append/conflict/byte-preservation since PR fix(weave): append-only current-only ReferenceCatalog inventory weave #34.
• 05-19_2349-branch-based-workingfile-fix claimed leakage validation, docs, and the SFLO replay were outstanding. All three landed; the replay ran 2026-08-01 with zero publication issues.

The carve

07-21_1603-extractor-defect-pair came back SHIPPED — correctly, on its own terms (F1 landed, F2 withdrawn by amendment r1 with no code deliverable). But it was also holding queue item 1. Closing it would have silently dropped a live epic.

Ruled in session: carve first. wa.task.2026.2026-08-02_1330-extracted-term-weave-batch-path now owns the scale work with the probe evidence — N=1,700 passes at 3.79 GiB against a ~4.09 GiB ceiling, because extracted candidates classify as firstExtractedKnopWeave and never enter the batch path PR #33 built (payloadBatchCandidates=0, 1,445,850 candidate-cache hits). The queue entry was repointed, never renamed, per the 2026-08-01 lesson that a closure sweep must exclude wd.queues. Gate green, 3 entries.

Still open for you

Three notes came back UNCLEAR — they need product decisions, not code evidence:

• 2026-05-04-fingerprint-verification — is an operator-facing verify surface still wanted? The digest substrate exists.
• 2026-05-25-markdown-it — retain or replace the hand-written renderer?
• 2026-06-12-rdfa-and-jsonld-support — "optionally embed" isn't classifiable scope.

🤖 Generated with Claude Code

[coderabbitai]

Important

Review skipped

Auto reviews are disabled on base/target branches other than the default branch.

Please check the settings in the CodeRabbit UI or the .coderabbit.yaml file in this repository. To trigger a single review, invoke the @coderabbitai review command.

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro Plus

Run ID: f9f046fd-6e2b-4db0-b7b5-f8c98f9dca7c

You can disable this status message by setting the reviews.review_status to false in the CodeRabbit configuration file.

Use the checkbox below for a quick retry:

• 🔍 Trigger review

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ All tests successful. No failed tests found.

📢 Thoughts on this report? Let us know!