CI-13 [HUMAN GATE]: Private-repository cutover — prove private-mode CI, budgets, runner trust; maintainer flips visibility for v0.3.0

[Chris0Jeky]

Wave: Smart CI Fabric (ADR-0066 acceptance conditions). Tracker: CI-00 #2324. Human gate: the maintainer performs visibility, billing, branch-protection, credential, package-visibility, publication, and runner-association actions. Agents prepare implementation and evidence; they never infer those actions complete.

Reconciled 2026-09-18 against the current release assessment in PR #3151. Historical comments remain part of the ruling/evidence record; newer explicit maintainer rulings and exact live settings remain authoritative.

Context and settled boundaries

Taskdeck's development repository goes private for v0.3.0 on a personal GitHub Pro account. The cutover must preserve a downloadable public release while keeping the development control plane private.

Settled boundaries:

• GitHub Actions has a $0 hard ceiling with stop-on-limit enabled.
• The public source/release mirror is Chris0Jeky/taskdeck-release.
• Release GHCR packages are made explicitly public and anonymously verified before repository privacy, then verified again afterward.
• sha_pinning_required: true is already enabled and verified; it is not a future cutover step.
• Smart CI / Required Gate is registered as required only after the repository is private, because a public fork can otherwise emit the same check context.
• Branch-current strictness, administrator enforcement, and break-glass are chosen from completed observation evidence, not pre-ruled here.
• Runner isolation, cleanup, offline, override, reset, and revocation are proved before any GitHub association.
• After privacy and before runner association, the complete private-mode rehearsal is fail-closed and Linux-only across every required, called, and reusable workflow. CI-17 CI-17: Add a fail-closed Linux-only private-cutover rehearsal across all workflows #3170 owns that mechanism.
• Full Windows evidence and the final exact-tag Windows archive use the approved isolated runner classes only after association, unless the maintainer explicitly re-rules the budget.
• The qualified private Release is published before the public mirror consumes and republishes it.

A. Preconditions before the visibility change

All evidence is bound to exact commits, workflow/config identities, and current live settings.

• Remaining v0.3 issues have an explicit SHIP, CLOSE ON EVIDENCE, EXPLICIT RESIDUAL, DEFER, or HUMAN GATE disposition; [Backend][Frontend][Review] Make proposal provenance metadata lifecycle- and authorization-safe #2315 retains its existing non-blocking residual ruling.
• CI-02/CI-03 planner and gate evidence meets the accepted observation window: at least 20 usable merged PRs after the last relevant fix, zero false reds, complete recall for enabled lane families, bounded normal merges, and fail-closed full escalation for direct pushes, missing receipts, moved bases, and ambiguous evidence.
• Landed verification has authoritative evidence collection and workflow integration, not only a decision core.
• CI-09 CI-09: Bounded CI caches and artifact retention with measured utility (storage under the Pro allowance) #2333 has a fresh identity-bound storage dry run, exact deletion authorization, execution ledger, and post-cleanup measurement below the private allowance under $0.
• CI-10 CI-10: One change-driven nightly coordinator, weekly full sweep, clean-from-tag release qualification #2334 has accepted nightly observation, a weekly full sweep, and the clean exact-tag qualification contract.
• CI-11 CI-11: Full-SHA action pinning, least privilege, and hosted-only qualification of CI-control changes #2335 and PR Disable checkout credential persistence and scope Pages permissions #2838 complete current-head least-privilege qualification and the ADR-0066/J.3 maintainer gate; the CodeQL posture is recorded.
• CI-16 CI-16: Public release/source mirror for the private development repository (SC-8 ruling 2026-09-03) #2439 has the approved mirror repository, narrowly scoped credential, fail-closed publishing workflow, and a no-publish rehearsal. Mirror Actions remain disabled.
• CI-04 CI-04: Isolated on-demand self-hosted Linux/Windows runners in personal-account mode (no secrets, offline by default) #2328 proves isolated VMs, no host mounts or personal credentials, one job per host, hosted override, offline behaviour, workspace/Docker/cache cleanup, VM reset, and revocation while runners remain unassociated.
• CI-17 CI-17: Add a fail-closed Linux-only private-cutover rehearsal across all workflows #3170 implements and proves the trusted Linux-only private rehearsal control across required CI and every called/reusable workflow. Contract tests discover every Windows selector and fail if any lane can bypass the control.
• Public-mode dry runs cover release, mirror, GHCR, private-rehearsal selection, and all expected Linux/control/security evidence.
• Public assets, Pages, packages, Releases, install links, licensing, support, telemetry, launch-kit, and awesome-selfhosted references have an explicit post-cutover destination.
• Codex/Copilot/private-mode review and Actions billing behaviour is verified rather than assumed; review cadence runs after CI stabilises instead of after every micro-push.

B. Maintainer cutover actions, in order

Each action is recorded here with date, actor, exact setting or command, and evidence. Stop on any mismatch.

1. Pause merges and select a short frozen cutover window.
2. Capture current branch protection/rulesets, Actions permissions, package visibility, Pages, Releases, collaborators, forks, external surfaces, and rollback values.
3. Review and authorize only the fresh identity-bound storage deletion set from CI-09: Bounded CI caches and artifact retention with measured utility (storage under the Pro allowance) #2333; execute and remeasure.
4. Create/verify Chris0Jeky/taskdeck-release, keep mirror Actions disabled, and install the narrowly scoped publishing credential without exposing it in issues, logs, or repository files.
5. Set every release GHCR package explicitly public and prove anonymous pull/read access.
6. Confirm the Pro plan and the settled $0 stop-on-limit posture.
7. Change the development repository visibility to private. Agents never perform this action.
8. Immediately register Smart CI / Required Gate, retain the security contexts, and apply the evidence-based strict, administrator-enforcement, and break-glass decisions.
9. Re-check Actions permissions, fork approval policy, Dependabot, Pages, packages, Releases, mirror links, collaborators, forks, and anonymous GHCR access.
10. Keep every self-hosted runner unassociated. Run R0/R2/R4 PRs, one normal merge, one nightly dispatch, and one no-publish release rehearsal through CI-17's trusted Linux-only mode. Abort the cutover if any private hosted Windows job is scheduled, the control is absent/bypassable, or expected Linux/control/security evidence is missing.
11. Only after step 10 succeeds, associate the already-proven isolated runners if hybrid mode remains desired. Verify labels, read-only token posture, no secret exposure, selected self-hosted workload evidence, and full Windows evidence on approved runner classes.
12. If hybrid mode is not retained, record hosted-only operation and explicitly re-rule the final Windows build budget/path before creating the release tag.
13. Record the complete evidence ledger on CI-00 CI-00 [TRACKER][decision]: Smart CI Fabric and private-repository readiness (ADR-0066) — go private for v0.3.0 on a personal GitHub Pro account #2324 and resume merges only when every cutover assertion is reconciled.

C. Final release actions after cutover

• Freeze one exact main commit and stop feature merges.
• Complete final-head checks and a no-publish rehearsal.
• Activate and prove a draft-only hold so the tag-triggered workflow cannot publish before post-tag evidence is accepted.
• Create the real v0.3.0 tag on the frozen commit.
• Qualify that exact tag on hosted Linux/control lanes and the secret-safe isolated Windows release runner.
• Reconcile both runner-class evidence sets to the same tag, commit, policy, checksums, provenance, and release contract.
• Consumer-smoke the Windows archive, immutable container, MCP proposal flow, upgrade path, and supported backup/restore claims.
• Publish the private Release with the exact qualified assets, checksums, provenance, and body.
• Let the mirror stage and verify its source snapshot and byte-identical assets before publishing its public commit, tag, and Release.
• Verify mirror downloads, links, checksums, provenance, and GHCR access anonymously before announcement.

D. Rollback and abort rules

• Keep previous workflow/config files reachable by immutable commit and capture all previous setting values before mutation.
• The hosted control override remains available; runners can be detached and revoked in one step.
• Do not flip public merely to regain free minutes. Diagnose against the captured state and use the documented override/rollback path.
• Do not create the release tag while the publication hold, runner boundary, Linux-only rehearsal, public GHCR continuity, mirror handoff, or exact-head evidence is unproved.
• A failed or ambiguous assertion stops the cutover. Record the failure and restore the captured safe setting where applicable.

Acceptance

• Sections A and B are complete with exact evidence.
• The development repository is private.
• Smart CI / Required Gate and retained security checks are enforced under the recorded policy.
• The pre-association private rehearsal used CI-17's fail-closed Linux-only mode and scheduled zero private hosted Windows jobs.
• Any associated runners were already proven and passed their separate post-association workload evidence.
• Public GHCR access, Pages, mirror source, Releases, assets, checksums, provenance, and links survive the cutover.
• The real tag can be qualified and published through the approved runner and mirror boundaries without exceeding the settled budget.

[Chris0Jeky]

Two gaps in the A-to-I prerequisite set, found while building a standing v0.3.0 readiness view (PR #2438). Both are answerable in one reply; neither is urgent.

SC-6 in OUTSTANDING_TASKS.md permits the visibility change only after sections A to I of docs/ci/PRIVATE_REPO_CUTOVER_CHECKLIST.md are complete. Mapping those sections to their named owners against the live milestones gives this:

Section	Owner	State
A. Decisions (maintainer)	#2324, #2337	Human
B. Measure before changing	CI-01 #2325, CI-09 #2333	#2325 closed; #2333 open, v0.3
C. Planner and gate	CI-02 #2326, CI-03 #2327	Open, v0.3
D. Event topology	CI-03 #2327	Open, v0.3
E. Test right-sizing	CI-05 #2329, CI-07 #2331, CI-08 #2332	Open, v0.3
F. Runners	CI-04 #2328	Open, v0.3, human
G. Supply chain	CI-11 #2335	Open, v0.3
H. Nightly and release	CI-10 #2334	Open, v0.4
I. Rehearsal while still public	none named	unowned

Everything lines up except the last two rows.

Q1. Is checklist section H really a cutover prerequisite?

#2334 (CI-10: change-driven nightly coordinator, weekly full sweep, clean-from-tag release qualification) sits on the v0.4 milestone, but SC-6 makes section H a prerequisite for a v0.3.0 action. Those cannot both be right.

• A. Section H is genuinely required before cutover. Then #2334 moves to v0.3 and becomes a release blocker.
• B. Section H is not required before cutover. Then the checklist should say so, and SC-6's "sections A–I" narrows to the sections that are.
• C. Part of section H is required and part is not. Then it needs splitting at whatever boundary you name.

Q2. Who owns checklist section I?

Section I lists eleven pre-cutover rehearsal scenarios (R0/R2/R3/R4 PR classes, a cancelled/superseded PR, a normal merge to main verifier, a direct-push escalation, a nightly no-change skip, a weekly deep run, and a release dry-run with no publish), all hosted-only with no runner attached. No issue owns them, so nothing tracks whether they have been run.

• A. Seed a CI-14-style child issue under #2324 that owns the rehearsal matrix and its evidence.
• B. Fold it into #2327, which already owns the gate contract and event topology.
• C. Keep it as a checklist-only item you run yourself at cutover time, with no issue.

Not asserted

I have not changed any milestone, added any issue, or edited the checklist. The readiness view in PR #2438 records both rows as open questions rather than resolving them, so whichever way this goes the file follows the answer instead of pre-empting it.

[Chris0Jeky]

Maintainer decision packet applied (2026-09-03). The packet is a ruling record from the maintainer's decision cockpit, not evidence that any external action was performed; docs and OUTSTANDING_TASKS.md are updated in PR #2442.

Rulings for the CI-13 human gate:

• SC3 — Actions overage ceiling: $10/month hard ceiling. Applying it (Billing → Spending limits), confirming the Pro plan and verifying Codex/Copilot billing on a private repository remain the maintainer's own actions; none is inferred as done. Section A checkbox stays open until the setting is evidenced here (B2).
• SC8 — public-asset decision: private development repository + public release/source mirror. Releases, checksums/provenance and the GPL source stay public through a mirror; GitHub Pages keeps publishing; the launch kit and awesome-selfhosted wording point at the mirror. Mechanics seeded as CI-16 #2439; docs/ci/PRIVATE_REPO_CUTOVER_CHECKLIST.md A and docs/REVIVAL_PLAN.md commitment 6 record it.
• SC1 confirmed with the private-Pro approval-boundary amendment (ADR-0066; step J.6 re-checks the Actions approval policy in private mode).
• SC4 — no pre-ruling on strict / enforce_admins / break-glass; decided after the observation window.
• SC-5 (sha_pinning_required), SC-6 (visibility) and SC-7 (runners) stay blocked on their prerequisites.

[Chris0Jeky]

Correction to my earlier packet: Q2 was wrong, please disregard it. Q1 stands.

I claimed checklist section I had no owner. It does. Line 3 of docs/ci/PRIVATE_REPO_CUTOVER_CHECKLIST.md reads "Executable copy: CI-13 #2337 (record evidence there)", which covers section I exactly as it covers every other section. There is nothing to assign and no decision needed. Caught in review of PR #2438 and corrected there.

A second correction from the same review, in the opposite direction, which matters more:

Section F is not a human gate, and I had it in the wrong column. Its four boxes are isolated VMs, no host mounts or personal credentials with one job per host, a tested hosted override and offline-runner behaviour, and tested workspace/Docker/cache cleanup with a documented VM reset and revocation path. That is all agent-preparable and all of it must land before cutover. Only the registration tokens and the GitHub association are human, and those are SC-7, which runs after. Classifying #2328 as wholly human would send required pre-cutover engineering out of the technical queue and let SC-6 look ready while section F is unbuilt.

Q1 is unchanged and still needs your answer: CI-10 #2334 owns checklist section H while sitting on the v0.4 milestone, and SC-6 makes sections A to I prerequisites for a v0.3.0 action. Either section H is not truly required before cutover, or #2334 belongs in v0.3.

Noted separately: I have not touched OUTSTANDING_TASKS.md for the SC-3 and SC-8 rulings. PR #2442 owns that file and is applying the whole 2026-09-03 packet, so recording them from a second PR would put two writers on one seam.

[Chris0Jeky]

Q1: agent recommendation, not a ruling (coordinator pass, 2026-09-03).

The inconsistency: docs/ci/PRIVATE_REPO_CUTOVER_CHECKLIST.md section H is owned by CI-10 #2334, which sits on the v0.4 milestone, while SC-6 makes sections A to I a v0.3.0 prerequisite. Section H has three boxes, and they are not the same kind of work:

1. One coordinator owns nightly/quality; honest-green no-change night; weekly full sweep. Consolidation. Nothing in private-mode cutover depends on it; the nightlies keep running in private mode as they do today, only against the private minute allowance. docs/ci/SMART_CI.md §9 already places nightly consolidation in phase 3, behind a 100 percent recall gate that itself needs the SC-4 observation window to complete.
2. Mutation remains manual (ADR-0052). Already true since 2026-08-19; a checkbox to tick, not work.
3. Release rebuilds from the exact tag in a clean hosted context; SBOM/provenance and digest verification release-owned; a release dry-run succeeds with ordinary self-hosted runners offline. This is the only box cutover depends on: v0.3.0 is cut from the private repository, so the tag-driven release must be proven to run hosted-only with no runner attached. Its concrete residual today is the double trigger in ci-release.yml and release-security.yml (push: tags: v* plus release: published), already recorded on #2334.

Recommendation: option B, split the section rather than move the issue. Keep #2334 on v0.4 for boxes 1 and 2 (rename its section H ownership to "H.1 nightly consolidation"), and carve box 3 out as a small v0.3 child (CI-10a "clean-from-tag release qualification, hosted-only dry run", owner to be a Codex lane) that becomes the section H prerequisite SC-6 actually needs. CI-09 #2333 (section B) already owns the release cache and artifact boundaries, so CI-10a is a natural sibling of the two open #2333 PRs.

Why not the alternatives:

• A: move #2334 wholesale to v0.3. Pulls nightly consolidation, which is gated behind recall evidence the SC-4 window has not produced, onto the release-critical path. That makes v0.3.0 wait on v0.4-class work.
• C: drop section H from the A-to-I prerequisite. Leaves the private-mode release path unproven at the moment the first private release is cut, which is exactly the failure the checklist exists to prevent.

If you rule B, the agent side is: open the CI-10a child on v0.3, edit section H into H.1 (v0.4) and H.2 (v0.3), and update docs/releases/V0_3_0_READINESS.md table row H. No maintainer action beyond the ruling.

[Chris0Jeky]

Maintainer rulings, 2026-09-03 (in-session, later than the packet).

Q1 = A. Checklist section H is a cutover prerequisite in full. CI-10 #2334 moves from v0.4 to v0.3 (Priority I) and is a release blocker; no split, no CI-10a child. The maintainer accepted that this pulls work forward. #2334 depends on CI-01 (closed), CI-03 #2327 and CI-05 #2329, both already v0.3, so no other issue changes milestone. Its scope still triages #1210 and #2180 (no milestone).

SC-3 re-ruled: deferred, no paid overage. The packet's $10/month ceiling is superseded. GitHub Pro is confirmed; the included 3,000 minutes/month are the entire hosted budget and are spent on Linux jobs only. Windows (x2) and any macOS (x10) legs run locally (the laptop registered as the Windows runner under CI-04 #2328; agent-run proving checks until then) or carry a local fallback. GitHub's default $0 spending limit is the effective hard ceiling.

Still the maintainer's on this issue: at step J.7 read Billing → Spending limits back, confirm it shows $0, record it here; verify Codex/Copilot billing on a private repository. Checklist A box 1 is checked; box 2 and the Windows-runner box carry the ruling.

Recorded in OUTSTANDING_TASKS.md §J, docs/ci/PRIVATE_REPO_CUTOVER_CHECKLIST.md A/H, docs/releases/V0_3_0_READINESS.md and ADR-0066 (amendment) in the same PR.

[Chris0Jeky]

SC-3 read-back, measured 2026-09-04 (Phase A of the private cutover, read-only, from the maintainer's own billing pages):

• Plan: GitHub Pro (education-discounted, 2 years remaining). Included-usage alerts: On.
• Budgets: Actions $0 budget, "Stop usage: Yes"; Packages $0 / stop usage yes; Codespaces $0; Git LFS $0; AI credit SKUs $0. No paid overage is possible — this is exactly the deferred-SC-3 posture (Linux-only included minutes, $0 ceiling).
• Consequence that changes the cutover order: active Actions artifact storage is 98.8 GiB (18,742 unexpired artifacts; 546 container-image-artifacts from main runs = 85.9 GiB; frontend-unit artifacts 7.4 GiB; perf/load harness ~4 GiB). Public-repository artifacts are free; on a private repository they bill per GB-day against the Pro allowance, and with a $0 Actions budget set to stop usage GitHub disables Actions the moment storage exceeds the included amount. So the flip with today's storage would stop CI on the private repository. Dry runs prepared (nothing deleted): --name-prefix container-image-artifacts --older-than-days 1 --include-protected → 549 artifacts = 92.8 GB; --all --older-than-days 1 --include-protected → 16,426 artifacts = 105.8 GB. The main/tag exclusion that SC-2 respected by design is what is left; deleting it is a maintainer authorization, tracked on the cutover readiness block. The five release-tag artifact sets (v0.1.0 … v0.3.0-rc.1, ~0.37 GB each) can be kept and still fit.

Still human after this: the Codex/Copilot billing behaviour on a private repository (Codex review credits are already exhausted — SC-9).

[Chris0Jeky]

Maintainer ruling, recorded 2026-09-06 (guided walkthrough, in-session interactive reply; map:v1:e5beef60c3235e76726721636988f12e53ad55368f875b9a64c8ce71ae621b93; OUTSTANDING_TASKS.md §J/§K).

Four §J items touched.

• SC-3 J.7 (q-5 = A, ticked). Billing → Budgets shows a $0 Actions budget, the effective hard ceiling under the deferred-overage ruling of 2026-09-03. The Codex connector bills through the maintainer's OpenAI subscription (allowance refreshes on that subscription's daily cycle; it reaches private owned repositories), so no GitHub-side billing applies. Copilot is the Student offer (134/200 included AI credits at read time, additional usage not enabled, $0 budget) and is not relied on.
• SC-9 (q-4 = A, ticked). Fresh-context agent review is the documentation-only gate whenever the connector is out of credits; no top-up. Codex resumes as primary when its allowance returns (maintainer: 2026-09-07).
• SC-10 (q-1 = A; q-2 = A). All twelve parked control-plane PRs are delegated to the coordinator: update-branch, fresh hosted run, fresh-context re-review where main touched the PR's files since its last review, merge in dependency order (#2506, #2502, #2522 then #2531 retargeted, then the rest); maintainer reviews post hoc. The five post-hoc merges #2479, #2529, #2548, #2549, #2556 are acknowledged, no revert.
• SC-11 (q-3 = A). The maintainer flips delete_branch_on_merge; the one-time sweep of merged-PR heads is authorized (ancestor-proven tips only, open-PR heads and bases excluded, everything else listed). Sweep evidence will be posted here.
• New item from D-14 (8): GHCR image visibility is configured separately from repository visibility; decide before SC-6 whether release-container.yml's images stay public when the repository goes private.

[Chris0Jeky]

SC-11 sweep executed 2026-09-06 under the maintainer's guided-walkthrough ruling q-3 = A (OUTSTANDING_TASKS.md §J SC-11).

Method: git ls-remote --heads origin (573 heads at classification time; 561 at the first count an hour earlier, the difference being branches a Codex session pushed in between) joined to the full PR inventory (1,538 PRs, open list refreshed immediately before deletion); a branch was deleted only if it was the head of a MERGED PR, was not the head or the base of any open PR, and its tip was proven an ancestor of origin/main (git merge-base --is-ancestor) both at classification and again immediately before its batch. Deletion went through git push origin --delete in 12 batches (no REST budget spent).

Result: 465 deleted, 0 failed, 0 skipped at re-verification. origin/main unchanged. Remote heads after: 108.

Left in place for a separate decision (not deleted, as the ruling requires):

45 merged-PR heads whose tip is NOT an ancestor of main (a later commit landed on the branch after merge)

• chore/harness-blueprint-step1 bbb20be
• docs/1405-codex-p2-fixes bb435ec
• docs/overnight-loop-refresh 4b8c9f4
• docs/post-merge-1195-1193-1131 25d4a60
• docs/post-merge-overnight-0717 36fb655
• docs/post-merge-overnight-0718 51dcc56
• feat/1195-bound-llmqueue-getbystatus be20515
• feat/1237-bound-pending-reads 05182be
• fix/1123-release-dry-run 27f4992
• fix/1164-serialize-cross-process-migrate 15df591
• fix/1377-deterministic-collision f7e1aaa
• fix/redirect-content-and-header-safety c9d050a
• issue-1274/paper-e2e-coverage 03ca77f
• issue-1282/sqlite-interceptor 1a0d6a4
• issue-1298/fake-undo-refresh 7660d96
• issue-1332/redis-connect-seam c911991
• issue-1335/hosted-worker-isolation 22e8a0e
• issue-1338/http-mcp-auth 202736f
• issue-1347/paper-review-enum-contract fce521b
• issue-1348/similar-past-sqlite 6e1dbe7
• issue-1355/export-blob-batching 9e3aab9
• issue-1357/migrator-test-hardening fea23c2
• issue-1365/presence-stabilization 8cb3da3
• issue-1367/allowedhosts-failclosed 0c39fe8
• issue-1368/preauth-failure-budget 4a91899
• issue-1369/pdf-extraction-timeout 9d4ebbf
• issue-1370/preview-apply-parity d2dcc88
• issue-1376/preview-apply-gates b954525
• issue-1382/dialog-handler-hardening c6edd8f
• issue-1383/audit-retention-hardening 6e6f0f6
• issue-1384/preauth-quota-gate 0df118c
• issue-1387/export-history-batch f0b36e2
• issue-1393/audit-culture-invariant ed65720
• issue-1402/lastused-persist 1b61e53
• paper/1004-today-dossier-api-wiring efb9663
• paper/1007-narrow-companions-complete 7697d38
• rfai-03/975-proposal-generator-v1 f5c3368
• rfai-04/976-edit-before-approve 01a6bec
• rfai-05/977-confidence-frontend 60483b0
• rfai-08/980-egress-registry af03611
• rfai-08/980-privacy-insights d055bee
• sec/1063-1064-security-fixes d0b734a
• sec/1065-health-info-disclosure 49dc969
• tst/1066-connector-abuse-tests 14e17da
• tst/1066-critical-test-coverage 3d833cc

27 heads of closed-unmerged PRs

• docs/sweep-2026-07-25-pr-queue 2238ca1
• fix/1293-floor-overlay 0c31da0
• fix/1531-review-doctrine-gaps f200013
• harness/tailor-2026-07-27 2f09394
• issue-1306/openai-compatible 6ee17a4
• issue-1323/prompt-rails-v2 7e65593
• issue-1379/extraction-resource-ceiling 7e7f36d
• issue-1456/codex-deny-floor-adapter-v2 6a09f43
• issue-1510/workflow-lint-clean 2114cc0
• issue-1547/atomic-evidence 8753063
• issue-1547/evidence-successor 5c6e4bd
• issue-1547/evidence-successor-v2 10f19dd
• issue-1625/codex-path-trust 04b50b1
• issue-1690/adr0051-mirrors bcd8ca2
• issue-1724/frontend-audit-refresh ef502da
• issue-1875/dev-docs 94390d3
• issue-1906/identity-bound-clean-reset 6af008e
• issue-1966/visible-filter-scope-v2 f248506
• issue-1967/applied-proposal-detail a70f88a
• issue-2005/capture-metadata-contract ab1db61
• issue-2005/capture-metadata-recovery-v2 9ac1fb5
• issue-2080/archived-write-guard 4b68e9b
• issue-2106/provider-native-board-token-migration ecc70bc
• issue-2205/apply-rate-binding ee84f1b
• issue-2475/pwa-activation-race b51002b
• issue-2563/sequence-pairing 5f1e873
• issue-2572/condition-wait 05008ba

11 branches that never had a PR

• followup/1553-hidden-index-preservation 5677f21
• integration/v0.3.0 ad53254
• issue-1242/durable-config 560758e
• issue-1242/release-relocation-smoke 53840f8
• issue-1274/paper-e2e-axe 993f188
• issue-1433/validate-before-persist fefa1ec
• issue-1876/desktop-retired-provider-diagnostics f010cb7
• issue-1940/provenance-shortcut c9135fe
• issue-1961/dedupe-board-fetch 08be057
• issue-2198/batch-approve-focus 9a2d723
• issue-2326/smart-ci-recall-report 3b29191

Also excluded by rule: 23 open-PR heads and the one open-PR base (issue-1898/port-release-deadline, base of #2531).

Still the maintainer's: flip delete_branch_on_merge (Settings → General → Pull Requests, or gh api -X PATCH repos/Chris0Jeky/Taskdeck -F delete_branch_on_merge=true). The SC-11 row closes when the setting reads true.