Decide where the work belongs
Map records at risk
For each job, record intended count, completed count, failed count, pending state, and downstream effect. Separate a read export from a mutation batch. Attach the operation handle and status evidence so a retry does not create a second job.
Compare recovery paths
Assess polling, result retrieval, per-record reconciliation, bounded resubmission, or owner review. Note retention and duplicate risks. Do not resubmit an accepted batch merely because progress is slow; first establish whether the original handle remains active.
Choose a job order
Prioritize partial mutations and expiring results, then blocked essential reads and ordinary delayed work. Set owner, review time, safe action, and evidence threshold. Keep a large but fully recoverable export below a smaller unknown write.
What to carry forward
Return a bulk queue with intended and completed work, operation age, partial-result risk, recovery path, owner, and review trigger. Escalate unknown mutations and expiring evidence. Keep ranking provisional when result retention or status semantics are unclear.
Technical background: GitHub documentation.
Keep the decision with the work.
Use a Work Item in Aglet to record the problem, the evidence you have, and the next decision. Add an owner and priority, then keep updates in the discussion so the next person can follow the reasoning.
Create an account See the product workflow