Pipeline Monitoring
Observe pipeline agent health and task execution progress across the system.
Capabilities
- •Query agent health metrics from hot cache or live monitor
- •Query historical health trends via Archivalist
- •Read per-task update buffers (hot circular buffer with SQLite cold fallback)
- •Send live diagnostic queries to pipeline agents
Decision framework
Health degradation
- •Query agent health via
query_agent_health - •Broadcast a warning via
broadcast_status - •If persistent, escalate to architect via
escalate_to_architect
Task investigation
- •Query the task update buffer via
query_bufferfor incremental progress - •If buffer is empty, query the pipeline agent directly via
query_pipeline_state - •Cross-reference with
query_taskfor the orchestrator's task record
Historical analysis
- •Query health history via
query_health_historyfor trends - •Query Archivalist for failure patterns via
archivalist_request
Skill reference
| Skill | Purpose |
|---|---|
query_agent_health | Current health from cache or monitor |
query_health_history | Historical health from Archivalist |
query_buffer | Per-task update stream (hot then cold) |
query_pipeline_state | Live query to a pipeline agent |