| name | whats-next |
| description | Analyze current benchmark state and recommend what to work on next. Triggers on whats next, what should I do, next steps, prioritize work. |
| user-invocable | true |
What's Next
Analyze the current state of benchmark runs and recommend the highest-value next action.
Steps
1. Get current status with gap analysis
cd ~/CodeScaleBench && python3 scripts/aggregate_status.py --gap-analysis --format json
2. Get config comparison
cd ~/CodeScaleBench && python3 scripts/compare_configs.py --format json
3. Categorize the situation and recommend
Based on the data, present recommendations using ALL applicable scenarios below. Always check for gaps first.
Priority 0: Missing runs (gap analysis)
If gap_analysis.total_missing > 0, this is the highest priority — we can't analyze what doesn't exist.
- Show total missing task runs vs expected
- Group by config (SG_full gaps are most critical since those are rerun-dependent)
- For each suite with gaps, show: suite name, config, count missing
- Suggest the appropriate
*_3config.sh script to run, or specific rerun commands
- Note: SG_full gaps are likely from archived DS-compromised runs that need rerun with the DS retry preamble
If runs are still in progress
Report:
- How many tasks are still running
- How many have completed so far (pass/fail/error)
