- Add stale task watchdog (requeueStaleTasks) to scheduler — detects in_progress tasks with offline agents and requeues or fails them - Fix Aegis rejection loop: rejected tasks now requeue to 'assigned' instead of staying in 'in_progress', with max 3 retries before failing - Track dispatch_attempts on tasks (migration 044) to prevent infinite retry loops — tasks fail after 5 dispatch attempts - Include error_message and reason in SSE event broadcasts so UI can show why a task reverted - Atomic task queue claim: replace SELECT-then-UPDATE race with single UPDATE...RETURNING statement - Gateway agent auto-registration: POST/PUT /api/gateways accepts optional 'agents' array to upsert agents (k8s sidecar support) - Document k8s sidecar deployment in docs/deployment.md Fixes: tasks stuck in assigned, Aegis rejection loops, agents invisible in k8s sidecar deployments |
||
|---|---|---|
| .. | ||
| plans | ||
| releases | ||
| LANDING-PAGE-HANDOFF.md | ||
| SCREENSHOT-GUIDE.md | ||
| SECURITY-HARDENING.md | ||
| cli-agent-control.md | ||
| cli-integration.md | ||
| deployment.md | ||
| mission-control-agents.png | ||
| mission-control-memory-graph.png | ||
| mission-control-overview.png | ||
| mission-control.jpg | ||