Triad migration evidence of 2026-10-07: Gate A refresh, the Gate B rehearsal with the merged adopter fix, the regenerated adoption plan, consumer-side reads and the Gate B checklist

[brettheap]

Part of #186

Gathers three finished evidence branches of the triad migration (feature 004-migrate-to-triad) into one pull request on top of main 837928a. Six files, all under specs/004-migrate-to-triad/ or openspec/changes/migrate-to-triad/, and nothing else: +2348 / -118.

What each branch adds, and who produced it

004-migrate-to-triad-exec: lane 1's Gate A writer.

• specs/004-migrate-to-triad/inventory.json refreshed: the 2026-10-07 inventory of 17 worktrees.
• specs/004-migrate-to-triad/preservation-2026-10-07.md, new: private backup ID prep-20261007T100559Z (host-local on Eagle); the source bundle holds all 33 refs and was restore-verified, with the same SHAs as the live listing and a clean git fsck --full.

004-migrate-to-triad-rehearsal: lane 1's Gate B writer.

• specs/004-migrate-to-triad/adopter-rehearsal-2026-10-07.md, new: the Gate B rehearsal with the merged adopter fix. Tool: openRepoShape 7f84ca4. Runs A and B both exit 0. All 100 paths accounted, 45 spec / 49 code / 6 root / 0 dropped. Legs: spec 558a12b, code 6d52162. Nested pin 7f84ca4, digest 3be52767….
• openspec/changes/migrate-to-triad/adoption-plan.yaml, REPLACED (regenerated, not appended). It now names source e43b243 (PR Bump the openRepoShape pin to 7f84ca4: #162's adopter fix for a source with an existing submodule, and #164's prefer-triad advisory #188's head, chore/openreposhape-pin-7f84ca4, since squashed as the repin on main) and carries execution_authorized: false. Its first line calls it "Rehearsal input, NOT the cutover plan".

004-migrate-to-triad-reads: lane openRepoTools-2 (additions only).

• specs/004-migrate-to-triad/consumer-reads-2026-10-07.md, new: what outside openRepoTools depends on its layout or its installer, 63 consumer rows, plus a Gate C follow-up list. Lane 2's six-check verification of it is on Triad migration (feature 004-migrate-to-triad): execution tracking under Brett Heap's word of 2026-10-07 #186.
• specs/004-migrate-to-triad/gate-b-checklist.md, new: the Gate B checklist walk, 46 checklist lines. It is a point-in-time walk (live state rechecked 2026-10-07T11:06Z): rows it labels "IN PROGRESS by lane 1's rehearsal writer" are answered by the rehearsal record in this same pull request, which it predates.

The other evidence branches (-installer, -ci, -root-guidance) are separate and are not in this pull request.

Authority

Nothing here authorizes real conversion. These are records of facts and a rehearsal; the plan file says execution_authorized: false, and all 19 tasks in tasks.md stay open. Gate C stays Brett Heap's explicit approval naming the actual revisions.

How the three branches were retargeted (no rebase, no force-push)

All three were cut from ff6f6a7, the head of PR #187's planning branch. #187 was squashed into main as 44983ce, so the 20 planning files are identical on both sides but GitHub's merge base is c1bac0d, and a plain merge reads a file a branch modified (adoption-plan.yaml, inventory.json) as "both added differently". Each branch B was therefore merged with the base pinned at ff6f6a7, so main's identical additions are a no-op (the method used for #175 and #179), against M = 837928a (main after #188):

T=$(git merge-tree --write-tree --merge-base=ff6f6a7 <B> 837928a)    # exit 0, no conflict
C=$(git commit-tree "$T" -p <B> -p 837928a -F <message>)
git push origin $C:refs/heads/<branch>                                # a fast-forward of the branch

Proof per branch: the +/- lines of git diff 837928a $C, with ^index and ^@@ lines stripped, are byte-identical (cmp) to those of git diff ff6f6a7 <B>.

Branch	Before	Retargeted head	merge-tree	Proof (stripped diff lines)
-exec	0fced47	e5f582a	exit 0	identical, 1523
-rehearsal	36c0324	4cdc06b	exit 0	identical, 536
-reads	d14b441	a6830a1	exit 0	identical, 702

Combined head 62ac10b (branch 004-migrate-to-triad-evidence): e5f582a + 4cdc06b -> e760f3c, then + a6830a1 -> 62ac10b, each step git merge-tree --write-tree exit 0 (the files differ; the tree is the same with the base left to git or pinned at main). Proof: git diff --name-only 837928a 62ac10b lists exactly the union of the six files the three branches touch; for each file git diff 837928a 62ac10b -- <file> equals git diff ff6f6a7 <that branch> -- <file> (stripped); and the whole stripped diff is 2761 lines, the sum of 1523 + 536 + 702.

The merge base of this head and main is 837928a, so the pull request shows only the six files. Land by squash.

🤖 Generated with Claude Code

Summary by Sourcery

Consolidate refreshed preservation records, the repinned adoption rehearsal, consumer dependency analysis, and Gate B checklist into a non-authorizing triad migration evidence set.

New Features:

• Add dated Gate A preservation, Gate B rehearsal, consumer-read, and checklist evidence for the triad migration.

Enhancements:

• Regenerate the adoption plan against the repinned source and record the current 100-path mapping, follow-ups, and non-authorizing rehearsal status.
• Refresh the migration inventory and document backup restoration verification for worktrees, refs, installed artifacts, and dependency pins.
• Document external consumers that depend on the repository layout or installer behavior and identify follow-up work required before Gate C.

Documentation:

• Add public migration evidence covering preservation, adopter rehearsal results, consumer dependencies, and Gate B status.

Tests:

• Record successful isolated adopter rehearsals, path/blob accounting, nested dependency verification, recursive bootstrap, and restore checks.

Chores:

• Keep all migration tasks open and explicitly preserve the prohibition on real conversion, repository creation, or Gate C authorization.

[sourcery-ai]

Sorry @brettheap, you've used your own review budget of 250,000 diff characters for the last 7 days.

You can request another review in 2 days and 2 hours by commenting @sourcery-ai review. Upgrade to get a review now.

[sourcery-ai]

Reviewer's Guide

Consolidates three retargeted evidence branches into six migration records: refreshed Gate A preservation, a regenerated non-authorizing adoption plan, successful local Gate B rehearsal evidence using the merged adopter fix, consumer-impact analysis, and a detailed open Gate B checklist. The PR changes documentation and migration metadata only; it authorizes no repository creation, conversion, lane rebinding, or Gate C execution.

Sequence diagram for the non-authorizing Gate B rehearsal

sequenceDiagram
    participant Plan as AdoptionPlan
    participant Tool as adopt-project.py
    participant Shape as openRepoShape
    participant Check as GateBChecklist

    Plan->>Tool: plan
    Tool->>Shape: openRepoShape
    Shape-->>Tool: 100 paths across spec, code, root
    Tool-->>Plan: regenerated adoption-plan.yaml
    Plan->>Tool: check
    Tool-->>Check: rehearsal evidence
    Check-->>Plan: execution_authorized false

Loading

Flow diagram for triad migration evidence gates

flowchart LR
    A[Gate A preservation] --> B[Regenerated adoption plan]
    B --> C[Gate B rehearsal]
    D[Consumer reads] --> E[Gate B checklist]
    C --> E
    E --> F[Gate C approval]
    F --> G[Real conversion]
    B -. execution_authorized false .-> H[No cutover]

Loading

File-Level Changes

Change	Details	Files
Refreshes Gate A inventory and publishes a restore-verified preservation receipt while explicitly keeping migration authorization and Gate A completion open.	Records the 17-worktree, 33-ref inventory at the October 7 breakpoint. Documents host-local source and dependency bundles, manifests, digests, and restore checks. Preserves ignored workflow configuration and installed-artifact receipt details for later rollback and freeze work.	specs/004-migrate-to-triad/inventory.json specs/004-migrate-to-triad/preservation-2026-10-07.md
Regenerates the adoption mapping from the repinned source and documents a successful Gate B extraction rehearsal with the merged openRepoShape fix.	Updates the plan to source commit e43b243 with 100 paths, 45 spec / 49 code / 6 root ownership, resolved new paths, follow-ups, and execution_authorized: false. Captures isolated local-remotes runs A and B, exact path/object accounting, kept histories, nested gitlink and digest validation, recursive bootstrap, and source immutability checks. Keeps Gate B open and identifies remaining root-guidance, installer, compatibility, CI, restore, and approval work.	openspec/changes/migrate-to-triad/adoption-plan.yaml specs/004-migrate-to-triad/adopter-rehearsal-2026-10-07.md
Adds consumer-side migration impact analysis and a point-in-time Gate B checklist to identify layout, installer, worktree, and external integration obligations.	Audits workBenches, openRepoShape, brett-wip, host state, installed artifacts, and openRepoTools assumptions across 63 consumer rows. Classifies uncovered dependencies and proposes 33 Gate C follow-ups with additive-before-cutover versus real-legs timing. Walks 46 Gate B requirements, distinguishing existing evidence, in-progress work, unowned work, and hosted-runner blockers without marking tasks complete.	specs/004-migrate-to-triad/consumer-reads-2026-10-07.md specs/004-migrate-to-triad/gate-b-checklist.md

Possibly linked issues

• Triad migration (feature 004-migrate-to-triad): execution tracking under Brett Heap's word of 2026-10-07 #186: The PR directly delivers the issue's requested Gate A refresh and Gate B rehearsal evidence under Part of #186.

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[brettheap]

Pushed 3f7c48f to fix the red hygiene gate (guard-launch-mode, tests, tests-no-submodule, tests-windows).

Cause. tests/test_repo_hygiene.py::test_no_committed_file_names_a_host_absolute_path refuses any /home/<user>/ path in a tracked file. specs/004-migrate-to-triad/consumer-reads-2026-10-07.md carried 24 of them (/home/brett/…). A scan of every tracked file with the test's own HOST_ABSOLUTE_PATH regex found no other file.

Change. In that one file, all 24 /home/brett/ are now ~/ (the document already uses ~/ for its other home paths): 19 in the recorded-commands block under "~/.agents and the home directory", 2 in the lanes/LANES.md:1109 row, 1 in the ~/.claude/settings.json symlink row, 2 in the "commands this revision added" block. One sentence under the title says why and that the originals are in lane 2's transcript. No row was deleted, renumbered or reworded. /workspace/… paths are untouched; the regex does not refuse them.

The document is lane openRepoTools-2's; this is a mechanical portability edit on lane 1's combined branch, made with the lane owner's word.

Verified. The test passes (-k no_committed_file_names_a_host_absolute_path: 1 passed, run under the workstation pytest flock), git diff --check is clean, and the same scan now finds no hit. Labels untouched; not merged.

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[brettheap]

Independent adversarial review — lane openRepoTools-3 (read-only; 2026-10-09T00:29:57Z)

Verdict: LAND, on head 3f7c48f3cd9f7e4572b1f671654307f9215e5244. The head moved during this review: the brief named 62ac10b, and 3f7c48f (pushed 2026-10-08T21:40:35Z) fixes the red hygiene gate. Both heads were reviewed. Every check below ran from a fresh scratch clone of opensoft/openRepoTools, with a scratch worktree at 3f7c48f, never from the author's trees. This PR is evidence-only: six files under the two feature directories, squash-clean onto 837928a. Every figure git or GitHub can check reproduces, and the one CI break (24 /home/brett/ paths) is fixed on 3f7c48f, where the Linux record is green. Two things stay with the landing owner (lane 1): the tests-macos gate on the ready label (in progress since 2026-10-08T22:22:08Z at this stamp), and the order against 004-migrate-to-triad-rehearsal-c (finding 4).

Findings, numbered. Finding 1 blocked 62ac10b and is fixed on 3f7c48f. None of the others blocks.

1. CI diagnosis. Red on 62ac10b from one test, fixed on 3f7c48f. In run 37842940685, all four red jobs fail exactly one test, tests/test_repo_hygiene.py::test_no_committed_file_names_a_host_absolute_path:

   • guard-launch-mode: 1 failed, 164 passed, 1131 deselected
   • tests: 1 failed, 1295 passed
   • tests-no-submodule: 1 failed, 1083 passed, 212 skipped
   • tests-windows: 1 failed, 154 passed, 1141 skipped

   The assertion fails at tests/test_repo_hygiene.py:2665 with AssertionError: committed file(s) name a host-absolute path: {'specs/004-migrate-to-triad/consumer-reads-2026-10-07.md': ['/home/brett/', …]}, listing 24 entries. The pattern is HOST_ABSOLUTE_PATH, alternative /home/[a-z][a-z0-9_-]*/ (tests/test_repo_hygiene.py:2259-2264).

   Cause. At 62ac10b, consumer-reads-2026-10-07.md names /home/brett/ 24 times: lines 146-158 (19 hits, the ~/.agents command block), 202 (2), 227 (1) and 391-392 (2). git blame puts 15 of those lines in ae70dc5 and 2 in d14b441. No other tracked file matches.

   Minimal fix. Replace /home/brett/ with ~/ in that one file. That is exactly what 3f7c48f does. I rebuilt 3f7c48f's file from 62ac10b's, with that substitution and the two-line note after the title, and it compared equal. No row was otherwise changed. A scan of the whole 3f7c48f tree with the test's regex finds 0 hits.

   CI on 3f7c48f: two complete runs, both green. guard-launch-mode passed at 21:41:02Z and 22:22:22Z, tests at 22:16:00Z and 22:55:58Z, tests-no-submodule at 22:21:13Z and 23:06:30Z, and tests-windows at 21:41:29Z and 22:22:35Z. parse-macos and SonarCloud passed. tests-macos was skipped in the first run and is in progress under ready.

   Local reproduction through tests/run.sh -k 'test_no_committed_file_names_a_host_absolute_path': both attempts (one per head, bounded at 3000 s each) timed out with exit 124 in run.sh's pre-lock census, because other repositories' pytest suites stayed live on the workstation throughout, so no assertion ran that way. As a substitute that adds no pytest load, I imported the test function from each tree's tests/ and called it directly. It raises the same AssertionError at 62ac10b (one file, 24 entries) and passes at 3f7c48f, and both trees stayed clean.

2. Scope and squash. All six files are under specs/004-migrate-to-triad/ or openspec/changes/migrate-to-triad/. Nothing else is touched by the net diff (+2350 / -118 at 3f7c48f; the body's +2348 is 62ac10b's).

   • Merge base is 837928a. git merge-tree --write-tree 837928a 3f7c48f exits 0, and its tree 96cc346 equals the head tree, so a squash lands exactly the head tree.
   • The two modified files start from main's own blobs (adoption-plan.yaml f9d26fb, inventory.json cb4dd82), which are also ff6f6a7's.
   • #187's 20 planning files are byte-identical between ff6f6a7 and 837928a. No path under either directory appears in git diff --stat ff6f6a7 837928a.
   • The body's retarget proof holds at blob level: every file at 62ac10b is blob-identical to its source head (0fced47 ×2, 36c0324 ×2, d14b441 ×2).
   • Nothing of #187 is lost except by intent: the October 4 plan is replaced (it stays at 44983ce in main's history), and see finding 6 for one inventory block.

3. adoption-plan.yaml is accurate for the source it names.

   • Source fields. e43b243 (#188's pre-squash head; c4864ac plus the pin and gitlink only): files 100, bytes 6139047 and commits 91 all reproduce.
   • Entries. All 34 entries' files/bytes match e43b243's tree. The summary 45/49/6/0 equals the per-entry sums.
   • Overrides. Comparing the parsed YAML, the 19 October 4 resolutions are carried verbatim: no leg changed, no resolution changed, no entry dropped.
   • The 14 paths that had no owner are now mapped, each with a resolution: ideation/ (12 files: README.md plus 11 under brainstorm/) to spec, and lane-worktrees and lanes-index to code.
   • Exactly one owner for every path at e43b243 (100), at main 837928a (120: spec 65, code 49, root 6) and at 3f7c48f (124). The 20 or 24 newer files fall under openspec/ and specs/.
   • Other fields. follow_ups has 29 entries: the tool's 19 plus the 10 reasoned ones, as the rehearsal record says. execution_authorized: false; planning_status: rehearsal-input-not-cutover.

4. This PR's plan conflicts with run C's. Decide the order before landing. 004-migrate-to-triad-rehearsal-c @ 5aa40b7 (cut from 837928a; #186 comment 6068938765) rewrites the same file, to source 837928a, 120 files, 65/49/6. git merge-tree --write-tree 3f7c48f 5aa40b7 exits 1, so whichever lands second conflicts on adoption-plan.yaml.

   • After this squash, main's plan would name e43b243. That commit is not an ancestor of main; it survives only as refs/pull/188/head, since the branch is deleted.
   • Its rehearsed legs and split (558a12b, 6d52162, 2726d3a) were wiped with /tmp on 2026-10-08 (#186 comment 6068861167).
   • Options: land #189 and then take run C's file when resolving the conflict on -rehearsal-c, or drop adoption-plan.yaml from #189 and let run C carry the plan. Lane 1's call.

5. Stale references this PR creates on main are only partly disclosed. adopter-rehearsal-2026-10-07.md:181-184 says plan.md, quickstart.md and preparation.md "still describe the October 4 version". It leaves out:

   • openspec/changes/migrate-to-triad/design.md:44-46: "The resolved adoption plan accounts for all 64 baseline paths: 18 spec, 40 code, six root, none dropped, with 28 follow-ups … all 16 originally unresolved paths". After this PR, the linked file has 100 paths, 45/49/6, 29 follow-ups and 18 tool-unresolved.
   • design.md:5-6
   • proposal.md:78

   Name them in that sentence too, or leave them to the owner's refresh, named.

6. inventory.json is valid JSON and consistent with GitHub at its stamp.

   • Stamp and PRs. The stamp is 2026-10-07T10:09:03Z, with main c4864ac. open_prs is [#187], and #187 opened at 10:03:43Z while #188 opened at 10:10:43Z.
   • Merged lists. merged_since_2026_10_03 (18) and merged_since_previous_snapshot (10) equal GitHub's merged lists for those windows, number for number.
   • Arithmetic. Owners 4+7+1+4+1=17; classes 8+6+3=17; 6 previous + 12 added - 1 removed = 17.
   • Gate flags. gate_a_complete and gate_b_complete are false (:1115-1116, :1641-1642), and limitations (:815) discloses the moving estate.
   • History. Every top-level value of main's inventory survives, unchanged or under superseded_snapshot_2026_10_04, with one exception: latest_refresh (:911) is overwritten. Its October 4 17:40→21:24 delta (real_conversion_waits_for_prs: [97, 121], "Owners continue #97/#121") is kept only in history at 44983ce. Minor.
   • Live state. Live state differs, as a stamped file should: main is 837928a, and #187/#188 are merged.

7. preservation-2026-10-07.md agrees with inventory.json on every figure.

   • Shared figures: 17 worktrees, 33 refs, 50 tips (33 + HEAD + 16), 2902 objects, 414 manifest files, 29/29 installs, the four SHA256s, and the backup id. Receipt f0ea2df7 is as in #186 comment 6036669816.
   • Landed heads. These match the merged PRs' head OIDs: #121 f13d329, #134 11037f0, #169 d2f8051, #174 4ea84c8, #179 6c79121.
   • One mismatch. closeout-2 (:83, feat/lane-end-closeout-rest at ed36551) is not #175's head. #175's head is 5c05ab6; ed36551 is an intermediate #175 commit, 59 behind it. Line 31's "at the heads of merged PRs … #175" is therefore imprecise for that row. The class still holds.
   • Paths. No machine path; the backup root is a relative pattern.

8. adopter-rehearsal-2026-10-07.md: every git-checkable figure reproduces.

   • New paths. 36 since daed209 (ideation/ 12, lane-worktrees, lanes-index, 11 under openspec/changes/, 4 under specs/005-…, 7 tests) and 6 since f5444c5.
   • Pins and blobs. Pins are 39d5c98/a44c0165… at c4864ac and 7f84ca4/3be52767… at e43b243, identical on 837928a. .gitmodules blob is 8043147a.
   • Kept commits are consistent with path-limited history: spec 34; code 87 and 88 with --full-history.
   • Follow-up notes. By git grep: 9 files name docs/, the ninth tests/test_repo_hygiene.py; 3 name openspec/, including lane-handoff; 1 names specs/.
   • Upstream. openRepoShape main...7f84ca4 is still identical.
   • Agreement with #186. It agrees with comment 6036661785 and with lane 2's six checks (6036838832): receipt adopter-fix-20261007-iM20tv (brett-wip c311a78b), 45/49/6/0 of 100, legs 558a12b/48d98cd/6d52162, splits 9e5f934/2726d3a, digests.
   • Honest about what it did not do: logs and paths stay private, Gate B is open, and T006–T011 remain.
   • Two statements are stale at landing and unmarked. Lines 8-12 and 208-209 call #188 the "repin candidate" that "still follows the usual landing rules"; it landed as 837928a. The disposable repositories it cites no longer exist (finding 4), so the private receipt's logs are now the only record. One dated line would cover both.

9. gate-b-checklist.md: counts and cites reproduce.

   • Counts: EXISTS 4, DECIDED 1, IN PROGRESS 19 (14+3+2), NOBODY 21, BLOCKED 1, for 46.
   • Cites. Spot-checked cites hold: plan.md:140-142 and :161-163, tasks.md:162-165 and inventory.json:513 at ff6f6a7, and the tests.yml job lines 52/72/126/157/232/285 at c4864ac.
   • Point in time. It is an 11:06Z snapshot. Its IN PROGRESS rows are answered by files in this same PR, and lines 61-62, 153 and 222 show #188 open. The PR body says so; the file does not.
   • Line drift. Row 24 (:178) cites consumer-reads "lines 235–239" and "Line 241" @ ae70dc556. In the landed file that content sits at 239–243 and 245: +2 from d14b441's header line and +2 from 3f7c48f's note. The cite is commit-pinned, so not wrong, but a reader on main lands 4 lines off.

10. consumer-reads-2026-10-07.md: counts reproduce; /workspace/ paths remain.

    • Counts. 63 rows (15+8+12+13+15); not covered 33 (9+3+11+5+5); partial 7. The Gate C table has 33 rows, timing 11/11/11, and 12 outside-change rows. All agree with #186 comment 6036711853. d14b441 is additions only (85/0), as its message says.
    • Machine paths. After 3f7c48f there is no /home/<user>/, but 86 lines still carry /workspace/projects/…, mostly in the "Exact commands" blocks (:44-142, :385, :391-396). The test does not refuse them, and some rows rightly quote /workspace/projects/openRepoTools as what consumers assume.
    • The rule they cut against. plan.md:38-40 says "Machine paths, secrets/transcripts and backup contents are omitted", and gate-b-checklist.md:15-17 (same lane, same PR) leaves out the git -C host path for exactly that reason.
    • Suggestion, not a block. Spell the command blocks with <projects>/…, the way they already use <scratchpad> and <ors.git>.
    • Private content. brett-wip (private) and ~/.agents appear only as file:line and short phrases, with no record bodies or credentials; #186's own body names the same handoff files.
    • Secrets. None of the six files contains credentials or token material. The only hits for secret/token/password are "no path had a secret-like name" (preservation) and "secret_like_paths_skipped": 0 (inventory).
    • The new note (:3). It points to "lane 2's transcript" for the originals, which is private by plan.md:39. The 24 original paths also stay readable in this PR's public branch history (ae70dc5…62ac10b) after the squash.

11. Commits. There are 17 at 3f7c48f: 6 are #187's pre-squash commits (52ce64e..ff6f6a7), content-identical to main and a no-op under squash, and 5 are merges.

    • Added then removed: the 24 /home/brett/ paths (ae70dc5, d14b441 → 3f7c48f). This is fine under squash.
    • Message against the final tree. ae70dc5's body says "The plan of record (baseline daed209) leaves lane-worktrees, lanes-index and ideation/ on main unmapped". That is false of the final tree, where 36c0324's plan maps all three. 76dcb93's "IN PROGRESS"/"#188 open" counts are point-in-time.
    • Write the squash body fresh rather than concatenating commit bodies.

Not checked here:

• the private receipts (brett-wip c311a78b, f0ea2df7), their logs and SHA256s, and the backup and restore themselves;
• the kept-commit counts by an actual filter-repo run;
• the live worktree list (the inventory is stamped, and I did not enter the shared checkout or any lane's trees);
• macOS, which waits on ready;
• a tests/run.sh run of the hygiene test (finding 1: the census never cleared).

Review posted by lane openRepoTools-3, as announced on #186 (comment 6069483615). Lane 1 owns the landing order (#186).