Goal
The [Senate] CI: Quest engine — generate tasks from quests when queue is low recurring task (id: 80ffb77b) runs every 30 minutes and has been returning "no work this cycle" (no-op) for every observed run. At 336 estimated runs/week with zero output, it is the single largest slot waster in the SciDEX fleet. This task investigates why and fixes the underlying condition or reduces frequency.
Acceptance Criteria
☐ Root cause of persistent no-op identified (e.g., queue threshold miscalibrated, quest scan bug, all quests exhausted)
☐ Fix applied: either the task begins generating tasks when queue is genuinely low, OR frequency reduced to every-4h with clear documentation of why
☐ At minimum: change frequency from every-30-min to every-4h to save ~294 slot-runs/week immediately
Approach
Read the quest engine spec: docs/planning/specs/quest-engine-ci.md
Check current open task count: orchestra task list --project SciDEX --status open | wc -l
Identify the queue-low threshold: what count is "low enough" to trigger generation?
Check whether any quests have pending generation tasks (quest state in DB)
If threshold miscalibrated: update it to reflect actual queue capacity
If quest scan has a bug: fix the scan logic
If all quests are genuinely exhausted: document and lower frequency to every-4h
Update task frequency via orchestra CLI: orchestra task update --id 80ffb77b-8391-493c-8644-37086c8e2e3c --frequency every-4hDependencies
- Resource intelligence audit: docs/planning/specs/resource_intelligence_audit_2026_04_23.md
Work Log
(timestamped entries added during execution)