Why
#377 / #378 land the burn-down kit and the run archive. The archive contains 13 verified patch sets that have gone nowhere. Without a home, they rot: the base branches move, the patches stop applying, and the 36 adversarial verdicts that vetted them become worthless. This issue is that home.
What has to happen per patch set
Each one needs, in its own repo: an open issue, a claim through claim-ticket.yml (passkey ceremony — one per claim), then a PR naming that issue. pr-claim will not accept a cross-repo trailer for a private repo, so the issue must live in the PR's own repo.
| id |
repo |
state |
trellis-2 |
trellis |
3 patches — confirm nix flake lock first; lock entries were hand-computed |
gr-cowork-port |
guest-room |
3 patches — now tracked by guest-room#88 |
claude-box-245 |
prx |
1 patch — must not say Closes claude-box#245; closes request 3 only |
conformance-25 |
conformance |
1 patch — report-only until promoted |
gpr-10 |
gh-project-room |
1 patch — does not close #10; see gh-project-room#104 |
synoptic-1 |
synoptic |
1 patch — macos-14 CI run is the only real proof |
synoptic-4 |
synoptic |
2 patches |
prx-360 |
prx |
1 patch |
site-32 |
site + bounded.tools |
2 patches, two separate PRs in two repos |
mint-18 |
mint |
1 patch — docs only |
fds-1 |
front-desk-scheduler |
1 patch |
prx-270 |
prx |
1 patch — changeset included; opens red without it |
dev-contracts-2 |
dev-contracts |
no patch — withdrawn; owner decisions only |
The standing caveat
The producing sandbox blocked registry.npmjs.org and npm.jsr.io. Every green check inside those reports ran against vendored or stand-in dependencies, on touched files only. CI in the target repo is the authoritative gate — treat all 13 as strong drafts, not as merges waiting to happen.
The real constraint
Thirteen patch sets means thirteen issues and thirteen passkey ceremonies. That cost is the reason .github-private (see the companion issue on a scoped claim capability) matters more than grinding this list by hand. Landing these is worth doing; automating the gate is worth doing first.
Reference
Once #378 merges: burndown/runs/2026-09-03/ — per-item REPORT.md, result.json, verdict-A/B/final.json, and APPLY.md with the command table. Set RUN=burndown/runs/2026-09-03.
Why
#377/#378land the burn-down kit and the run archive. The archive contains 13 verified patch sets that have gone nowhere. Without a home, they rot: the base branches move, the patches stop applying, and the 36 adversarial verdicts that vetted them become worthless. This issue is that home.What has to happen per patch set
Each one needs, in its own repo: an open issue, a claim through
claim-ticket.yml(passkey ceremony — one per claim), then a PR naming that issue.pr-claimwill not accept a cross-repo trailer for a private repo, so the issue must live in the PR's own repo.trellis-2nix flake lockfirst; lock entries were hand-computedgr-cowork-portclaude-box-245Closes claude-box#245; closes request 3 onlyconformance-25gpr-10synoptic-1synoptic-4prx-360site-32mint-18fds-1prx-270dev-contracts-2The standing caveat
The producing sandbox blocked
registry.npmjs.organdnpm.jsr.io. Every green check inside those reports ran against vendored or stand-in dependencies, on touched files only. CI in the target repo is the authoritative gate — treat all 13 as strong drafts, not as merges waiting to happen.The real constraint
Thirteen patch sets means thirteen issues and thirteen passkey ceremonies. That cost is the reason
.github-private(see the companion issue on a scoped claim capability) matters more than grinding this list by hand. Landing these is worth doing; automating the gate is worth doing first.Reference
Once
#378merges:burndown/runs/2026-09-03/— per-itemREPORT.md,result.json,verdict-A/B/final.json, andAPPLY.mdwith the command table. SetRUN=burndown/runs/2026-09-03.