Context
Branch: hybrid-agent-cli
Wave: 1 (blocks thesis from being reclaimable)
Found by: polytest run 20260423-133619, hit on DOMPurify
Bug
After the lead policy-rejected a PR (because it included files outside the editable surface), the thesis remained in "claimed" state. No release was posted. The thesis is permanently stuck — no other node can claim it, and the current node won't retry it.
The contribute agent doesn't check for rejected PRs and release the corresponding thesis. The lead doesn't release claims when it rejects a PR. Neither side cleans up.
Evidence (DOMPurify)
Thesis #1 state after policy rejection of PR:
- Phase: claimed
- Active claims: still held by the node
- PR: closed (policy_rejection)
- No release comment posted
The thesis will remain stuck indefinitely until manually released with polyresearch admin release-claim.
Possible fixes
- The
decide command should post a release comment when it policy-rejects a PR
- Or: the contribute workflow should check for closed/rejected PRs at the start of each cycle and release the corresponding claims
- Or: the duties system should detect this state (claimed thesis with closed PR, no release) and surface it as a blocking duty
Context
Branch:
hybrid-agent-cliWave: 1 (blocks thesis from being reclaimable)
Found by: polytest run
20260423-133619, hit on DOMPurifyBug
After the lead policy-rejected a PR (because it included files outside the editable surface), the thesis remained in "claimed" state. No release was posted. The thesis is permanently stuck — no other node can claim it, and the current node won't retry it.
The contribute agent doesn't check for rejected PRs and release the corresponding thesis. The lead doesn't release claims when it rejects a PR. Neither side cleans up.
Evidence (DOMPurify)
Thesis #1 state after policy rejection of PR:
The thesis will remain stuck indefinitely until manually released with
polyresearch admin release-claim.Possible fixes
decidecommand should post a release comment when it policy-rejects a PR