Agent Queue Archive¶
Purpose:
- preserve completed queue history without bloating the active queue,
- keep doc/governance/Agent_Work_Queue.yaml small and actionable,
- retain validator-compatible task records for reference.
Rules:
1. doc/governance/Agent_Work_Queue.yaml is the only active queue.
2. Archive files are read-only history unless a deliberate restore is required.
3. New work must never be claimed from an archive file.
4. Archive files should preserve the same top-level schema as the active queue:
- version
- updated_at
- states
- roles
- tasks
5. The active queue should preferentially contain only actionable work:
- todo
- in_progress
- blocked
Done items should be moved to archive files once their completion metadata is preserved.
Exception:
- completed tasks that are still referenced by active depends_on edges may remain in the active queue as dependency anchors until the queue graph moves to a structured store or the active tasks are updated
6. The active queue is an execution surface, not the long-term historical ledger. Detailed completion history belongs in archive snapshots or a future structured task store.
Transition note:
- Until the queue system moves to a structured source of truth, keeping the live YAML focused on active items is the main defense against size, merge, and syntax fragility.
- Structured-store design reference: doc/governance/Agent_Queue_Structured_Store_v1.md
Current archive files:
- Agent_Work_Queue_2026-03-09.yaml
- Agent_Work_Queue_2026-03-12.yaml