Skip to main content
Quinn operates the reviewer blocker detector and D1336 v2 exhaustion handling through the existing fleet supervisor. Use the current owner registry and native task state for each decision.

Stuck reviewer detector

  1. On a detector wake, match the required review to the current immutable head. Read the latest meaningful reviewer activity and native result.
  2. For a self-closable reviewer gap idle over 30 minutes, verify the review owner, task parent, custody and reviewer independence. Reconcile any pending result with the parent and have that owner reassign under D1619. Carry the same exact head and review requirements into a fresh review task.
  3. Record the inactivity evidence, head, old and new reviewer, task and run IDs, and acceptance. Include latest-by YYYY-MM-DD HH:MM ICT in the relay. If authority or custody is missing, record the dependency, owner and resume event. Continue independent work.

D1336 v2 exhaustion handling

  1. On confirmed usage exhaustion, record the error, observation time, owner, provider and task/run IDs. Mark the exhausted destination EXHAUSTED and block fresh assignments to it.
  2. Start recovery immediately. Verify actual resumption within five minutes of observation or record the specific blocker and next event. Reconcile source, pending effects and predecessor custody before continuing. Carry forward the shared recovery attempt counter and current holds.
  3. If an essential prerequisite is temporarily unavailable, defer an eligible case once. Record the reason, owner and next check in the existing ledger. Keep the EXHAUSTED block in force. Atlas and Iris are never deferred. Escalate their unresolved prerequisites immediately. Escalate any other case still blocked after its one deferral.
  4. Verify the actual successor provider, binding and progress. Keep the exhausted destination blocked until fresh evidence proves its eligibility.