| name | feather-logs-observability |
| description | Use when an agent needs to diagnose Love2D behavior from Feather logs, observer values, assets, runtime messages, or telemetry. |
Feather Logs And Observability
Workflow
- Start with live session context:
- Read
feather://sessions, then feather://sessions/{id}/logs, observers, assets, and performance.
- Use refresh tools when the snapshot looks stale.
- Separate signal from noise:
- Group errors, warnings, repeated logs, observer changes, and recent plugin payloads by time.
- Check whether runtime suspend/resume or plugin settings affected data freshness.
- Ask for targeted observers when code changes are needed:
- Add or update lightweight observers around the suspected system.
- Avoid broad object dumps that will hide timing-sensitive behavior.
- Verify with a second snapshot after the fix or action.
Output
Summarize the timeline: what changed first, what the runtime reported, which observer or log confirms it, and the next concrete check.
References
- Read
references/workflow.md for evidence collection, timeline construction, observers, and stale-snapshot handling.