[Deicyde]

Stacked on #53. The base branch is claude/docs-status-2026-10, so this diff shows only the archive change. Merge #53 first, then retarget this PR to main.

Release-neutral: every changed path is under docs/ (per site/ops/brain_release_authority.py).

Problem

docs/ holds six dated Brain SQLite checkpoints, a PR-stack review guide, two session handoffs and a 140 KB reconciliation report, side by side with the current plan and runbooks, and nothing separates them:

• docs/BRAIN-SQLITE-CONTINUATION-2026-09-08.md and docs/BRAIN-SQLITE-HANDOFF-2026-09-09.md mark themselves "Historical checkpoint only".
• docs/BRAIN-SQLITE-HANDOFF-2026-09-19.md, -09-23.md and -09-26.md predate the 2026-10-04 promotion. They still say the selector returns 404, and they carry host paths and instructions that no longer apply.
• docs/BRAIN-SQLITE-PR-REVIEW.md is the review guide for the PR stack merged on 2026-09-21.
• docs/SESSION-HANDOFF-2026-07-04.md and -08-06.md are one-session handoffs. The first cites a memory file that doesn't exist.
• docs/BRAIN-STATUS.md is the 2026-07-04 reconciliation of the pre-build proposal.

Change

• Moves those nine files to docs/archive/ as git renames, with no content changes except link targets.
• Adds docs/archive/README.md: one line per file, and a pointer to the current docs (ROADMAP.md and the 2026-10-04 activation record).
• Re-points every relative Markdown link: inside the moved files (to ../ targets), and from docs/ROADMAP.md, docs/BRAIN-SQLITE-HANDOFF.md, docs/BRAIN-ACTIVATION-2026-10-04.md and docs/BRAIN-SQLITE-CURATION-DECISION.md to the new paths. Two live references that called the 2026-09-23 handoff "current" now name it by date.
• docs/BRAIN.md (the 2026-07-03 proposal) gets a "Historical" banner and stays in place because code comments and brain/SCHEMA.md cite it. docs/BRAIN-V2.md gets a "Partly historical" banner: its data model is superseded by docs/BRAIN-V3.md and its authority and release design by docs/BRAIN-ARCHITECTURE.plan.md, but its benchmark (the spec for bench/), ingestion-matrix and "Nightly brain sync" sections are still cited, so it stays in place too. docs/BRAIN-EDITS-ROADMAP.md already marks itself historical and is cited by migrations and Worker code, so it is unchanged.

How tested

• A script resolved every relative Markdown link in tracked *.md files that points to a moved file or sits inside one: none are broken.
• git grep for the old docs/<name> paths outside docs/archive/ finds nothing.
• site/ops/brain_release_authority.py reports every changed path as release-neutral.
• An independent adversarial review of the diff (links, leftover references, index accuracy, banner claims) found no blockers. Its two should-fix items (the BRAIN-V2 banner's scope, and the "current handoff" link text) are fixed.

Not changed (notes for Jack)

• CLAUDE.md and AGENTS.md still call docs/BRAIN-V2.md the Brain "design". Pointing them at docs/BRAIN-V3.md would make this PR release-affecting while AGENTS.md is classed that way, so it is left for a follow-up.
• brain/build_common.py:2156,3190 and catalog/fetch_math_graph.py:12 cite BRAIN.md:452 by line number. That citation was already off by two lines before this PR, and the banner shifts it by five more. Citing the section ("TheoremGraph / TheoremSearch … deep-dive") instead would be a code change outside this docs-only PR.
• Archived files keep their own stale "current" pointers (for example the 2026-09-08 and 2026-09-09 banners). The archive index says nothing there is current.

Risks

• External links to the old docs/... URLs on GitHub stop resolving at those paths. Git history keeps the files, and the archive index lists them.

What Jack must do

• Merge Docs: record the Brain activation and correct project status #53 first, then retarget this PR's base to main and merge.

[Deicyde]

Review status: merged (reached main with #53 in 5fb12e02)

Update 2026-10-15: #53 merged in 5fb12e02, which carried this PR's archive renames to main; docs/archive/ is on main now. Nothing else is outstanding here.

Original review (2026-10-14)

Independent adversarial review of 9cdd4526: a defect reviewer with skeptics, a design reviewer with a devil's advocate, and an arbiter.

Verdict: merge as written, right after #53. This is a docs-only, release-neutral cleanup: nine pure renames into docs/archive/, a short index, re-pointed links, and banners on BRAIN.md and BRAIN-V2.md, which stay put because code, bench/ and brain/SCHEMA.md cite them. The review found 0 broken relative links across tracked .md files, no code, test, workflow or live-site link that reads a moved path, and green CI. Nothing is required before merge.

Optional

• Link ROADMAP "Waiting on Jack" item 3 (docs/ROADMAP.md:26-31) to the source-policy paragraph in docs/archive/BRAIN-SQLITE-HANDOFF-2026-09-23.md:288-311. It is the only in-repo list of those open scope questions.
• docs/archive/README.md:20: say "Historical or partly historical docs". BRAIN-V2.md's own banner says "Partly historical".

Decisions for the maintainer

• Merge steps: merge Docs: record the Brain activation and correct project status #53 with a merge commit, then retarget Docs: archive superseded docs #60's base to main and check the diff still shows only these 16 files. delete_branch_on_merge is false, so GitHub will not retarget automatically, and skipping this step lands the archive on the stale stack branch instead of main.
• Whether to ask Docs: record the Brain activation and correct project status #53 to point "design docs/BRAIN-V2.md" at docs/BRAIN-V3.md, and to reword "the Brain checkpoint named by docs/BRAIN-SQLITE-HANDOFF.md" as checkpoint history. Neither blocks Docs: archive superseded docs #60.

Merge order: after #53 only; if #53 picks up commits touching the moved or re-pointed docs, rebase #60 and rerun the link check.

Posted by the WikiLean PR swarm (lead); updated 2026-10-15.

[Deicyde]

Status review (2026-10-14): ✅ Ready after #53

Checked mechanically: every relative link resolves, nothing outside docs/archive points at the old paths, no open PR other than #53 (whose references this PR already rewrites) references them, and every changed path is release-neutral.

Order (important)

• Merge Docs: record the Brain activation and correct project status #53 with a merge commit. If Docs: record the Brain activation and correct project status #53 is squash- or rebase-merged, this PR conflicts in docs/BRAIN-ACTIVATION-2026-10-04.md, docs/BRAIN-SQLITE-HANDOFF.md and docs/ROADMAP.md, because it rewrites lines Docs: record the Brain activation and correct project status #53 adds (reproduced).
• Then retarget this PR to main by hand before merging. Head branches aren't auto-deleted, so GitHub won't retarget it, and merging as-is would land it on claude/docs-status-2026-10.
• Docs: record the Brain activation and correct project status #53 still needs its pending fix (the 2026-10-04 attempt was reconciled on 2026-10-08). None of the files this PR moves carry that text, and its link edits in the three shared files are on other lines, so a fix added as new commits on Docs: record the Brain activation and correct project status #53 should carry through cleanly. If Docs: record the Brain activation and correct project status #53 is rebased or force-pushed for the fix instead, rebase this branch onto it before merging.

Nits

• The new BRAIN.md banner shifts the BRAIN.md:452 licence citation further out. It is also written into Brain data _meta by brain/build_common.py and cited in catalog/fetch_math_graph.py:12. The intended "Do NOT scale up storage…" bullet is now at line 459. The PR acknowledges this; consider citing a heading instead of a line number.
• The PR body says SESSION-HANDOFF-2026-07-04.md cites a memory file that doesn't exist. It exists in the local Claude memory store, just not in the repo.

---

_{🤖 Automated status pass by Claude Code over all 18 open PRs (#46–#63), posted for the repo owner. An independent verifier re-checked each finding. All 18 heads were also merged and tested together in a scratch tree; the only test failure that comes from combining them is #47 × #55 (three other pairs had mechanical text conflicts). Reviewed head: 9cdd4526.}