[20/36] Record final OpenCollection supervisor acceptance

[cjohnsto-nz]

---

This is part 3 of 21 in a stack made with GitButler:

•  21  [22/36] Release 0.8.0 with packaged PDF.js preview #54
•  20  [21/36] Complete OC-070 protocol surface polish #56
•  19  [20/36] Record final OpenCollection supervisor acceptance #53 <-- current
•  18  [19/36] Add OC-110 runtime authoring UX #52
•  17  [18/36] Add OC-080 runtime lifecycle and OC-110 task setup #51
•  16  [16/36] Record OC-050 OC-090 OC-100 supervisor audit #50
•  15  [15/36] Add OC-050 auth and transport support #49
•  14  [14/36] Add OC-100 request type UX #48
•  13  [13/36] Add OC-100 request type UX task #47
•  12  [12/36] Add OC-080 and OC-090 task tracks #46
•  11  [11/36] Record OC-010 through OC-040 supervisor audit #45
•  10  [10/36] Add OC-020 WebSocket support #44
•  9  [09/36] Add OC-010 GraphQL support #43
•  8  [08/36] Require demo server fixtures for parallel tracks #42
•  7  [07/36] Record OC-000 and OC-060 supervisor audit #41
•  6  [06/36] Add OC-060 schema round-trip validation #40
•  5  [05/36] Add OC-040 runtime scripting support #39
•  4  [04/36] Add OC-090 gRPC streaming support #38
•  3  [03/36] Add OC-030 gRPC unary protobuf support #37
•  2  [02/36] Add OC-000 protocol dispatch foundation #36
•  1  [01/36] Add OpenCollection agent plan #35

[APKiwi]

Verdict: approve. All nine source paths in the rewritten evidence table exist at this head, and the claims are consistent with what the stack actually contains at this point. The ledger's specific test counts and install evidence are unverifiable from the diff, normal for a ledger entry. Same note as the other audits: the acceptance is build/test verification, not a security pass, two significant issues in the accepted tracks surfaced later in review.

[APKiwi]

Re-review (whole-stack pass, per REVIEW_GUIDE.md)

This PR is the final acceptance record, so its accuracy is the whole point. Two problems.

• (high) The AGENT_PROGRESS.md OC-070 row this PR adds marks OC-070 done, "no blocking issues", with a full evidence trail (16 files / 203 tests including test/oc070SurfacePolish.test.ts, a packaged VSIX, a live smoke run). None of that code is on this branch or any ancestor. test/oc070SurfacePolish.test.ts and the OC-070 importer/copilot changes land entirely in [21/36] Complete OC-070 protocol surface polish #56, which is based on this PR's head (git merge-base --is-ancestor final-review oc-070 succeeds, the reverse fails). Per the merge order this PR is 20/36 and OC-070 is 21/36, so when this lands OC-070's code is not yet in main. The acceptance record precedes the code it certifies, and [21/36] Complete OC-070 protocol surface polish #56 does not re-touch this row, so this is the only record of that acceptance.
• (medium) The report certifies OC-000 through OC-110 as accepted with no blocking findings, but none of the four post-review fix commits are ancestors of this head. src/models/types.ts:672-675 still has the isHttpRequest discriminator bug that a3cae8e fixes (which this branch lacks). The blanket "no blocking findings" is stale against what the human review actually found in OC-000/OC-060/OC-090.

Whole-stack, acceptance-blocking (surfacing here since this is the acceptance PR):

• No single tree containing all 36 PRs plus the four fix commits (ae8c54c, 5b5164f, d34d834, a200076) has been built or tested. A restack is needed before this record is actually final.
• fix/open-collection-pr-rework (4631068, WebSocket binary payload validation) has no PR in the 36-step order and gets dropped if the landing procedure deletes stack branches as written. It needs its own PR or a fold into [10/36] Add OC-020 WebSocket support #44/[29/36] Add OC-140 WebSocket lifecycle UX #61.

Verdict: changes-needed. Defer this ledger update until after #56 lands, or caveat the OC-070 entry as "verified in a composite working tree, code lands in #56", and reconcile the fixes before signing final acceptance.

[cjohnsto-nz]

The review findings are valid, and they are now resolved with branch ownership and stack integrity made explicit.

• OC-070 is not implemented in PR [20/36] Record final OpenCollection supervisor acceptance #53's ancestry. It lands immediately afterward in PR [21/36] Complete OC-070 protocol surface polish #56 at 154414a. Commit ba51406 adds pr-53-acceptance-scope.md to correct PR [20/36] Record final OpenCollection supervisor acceptance #53's record: its OC-070 evidence came from a composite workspace containing PR [21/36] Complete OC-070 protocol surface polish #56 and was never evidence that the files existed at PR [20/36] Record final OpenCollection supervisor acceptance #53's head.
• The later findings were also valid. Their fixes remain on the branches that own the affected code: ae8c54c on PR [02/36] Add OC-000 protocol dispatch foundation #36, 5b5164f on PR [04/36] Add OC-090 gRPC streaming support #38, and d34d834 plus a200076 on PR [06/36] Add OC-060 schema round-trip validation #40.
• The historical “no blocking implementation findings” statement is now explicitly scoped as dated build/test acceptance. It was not a security review and does not supersede findings raised later.
• The 36-PR tip, all owning-branch review fixes through this PR, and the stack-tip rework were composed in one isolated tree. npm run compile passed, all 509 tests in 29 files passed, and npm run build passed.
• The cross-cutting rework now has a landing path: draft PR [37/37] Reconcile OpenCollection PR review fixes #72 targets PR [36/36] Fix request action first-click reliability #71's branch. It is not disposable branch-only work.

I did not rewrite the heavily edited AGENT_PROGRESS.md sections on PR #53. The pre-push merge simulator proved that doing so creates a conflict with descendant ledger edits. PR #53 therefore carries the additive correction document, while PR #72 carries the corrected final ledger row and current composite verification record.

Stack checks are complete: updated PR #52 → PR #53, PR #53 → PR #56, and PR #53 → the rework branch all simulate cleanly. GitHub reports PR #53, PR #56, and PR #72 as MERGEABLE/CLEAN with all checks passing.

[cjohnsto-nz]

Response to the second review:

The acceptance-order finding was valid and is corrected on this PR's current head, ba51406. The OC-070 row now states that implementation is carried by downstream PR #56 and that the evidence was produced from a composite workspace; it no longer implies the implementation is an ancestor of PR #53.

The formerly untracked rework branch now has its own landing PR, #72. I have recomposed the live stack again after the second review: all 37 immediate base/head edges merge cleanly and all 37 heads compose sequentially. On the exact composed tree, npm run build and npm run compile pass, all 31 test files / 539 tests pass, and demo validation passes 47/47 files.

The historical acceptance record remains explicitly dated and does not supersede later security review findings. No further change belongs on PR #53.