What problem does it solve?
It solves the problem of not knowing which experiments are still running, which ones have anomalies, and which ones have finished but still need collection, so you can confidently decide the next action.
Core Features & Use Cases
- Live experiment status monitoring: Scans
running experiments and checks local screen sessions or remote sessions to classify them as running, anomaly, or completed pending collect.
- Optional auto-collection for finished runs: With
--collect-ready, it automatically triggers /exp-run --collect for experiments whose sessions have ended but are not yet collected in the wiki.
- Pipeline-aware reporting and advancement: With
--pipeline <slug>, it reports pipeline progress and, with --auto-advance, advances to Stage 4 only when all pipeline experiments are completed.
Use case example: After a research pipeline runs for hours, use this skill to identify which experiments are stuck or crashed (anomalies) and which are done but not yet collected, then either collect them immediately or let the pipeline auto-advance when everything is complete.
Quick Start
Ask: Run /exp-status --collect-ready to automatically collect all completed experiments whose sessions have ended.
