# helix-stats > Show learning system health - insight counts, tag distribution, effectiveness, feedback loop status. - Author: Chris Kroenke - Repository: enzokro/tether - Version: 20260206005336 - Stars: 0 - Forks: 0 - Last Updated: 2026-02-06 - Source: https://github.com/enzokro/tether - Web: https://mule.run/skillshub/@@enzokro/tether~helix-stats:20260206005336 --- --- name: helix-stats description: Show learning system health - insight counts, tag distribution, effectiveness, feedback loop status. --- # System Health Display memory system statistics and learning loop status. ## Execution ```bash HELIX="$(cat .helix/plugin_root)" python3 "$HELIX/lib/memory/core.py" health ``` ## Output Includes - **status**: `HEALTHY` or `NEEDS_ATTENTION` - **total_insights**: Count of all insights - **by_tag**: Breakdown by tags (debugging, pattern, eval, etc.) - **effectiveness**: Average effectiveness of insights with feedback - **with_feedback**: Count of insights that have received feedback (use_count > 0) - **issues**: List of problems if any ## Example Output ```json { "status": "HEALTHY", "total_insights": 30, "by_tag": {"debugging": 6, "pattern": 10, "eval": 5, ...}, "effectiveness": 0.65, "with_feedback": 12, "issues": [] } ``` ## Interpreting Results | Metric | Healthy | Action if Unhealthy | |--------|---------|---------------------| | total_insights | > 0 | Run helix to build memories | | with_feedback | > 0 | Ensure feedback() called after builds | | effectiveness | > 0.5 | Prune low performers | | issues | empty | Address listed issues | ## Maintenance Commands ```bash HELIX="$(cat .helix/plugin_root)" # Decay dormant insights toward neutral (0.5) effectiveness python3 "$HELIX/lib/memory/core.py" decay --days 30 # Prune insights that consistently fail (effectiveness < 0.25 with 3+ uses) python3 "$HELIX/lib/memory/core.py" prune --threshold 0.25 --min-uses 3 ``` ## Feedback Mechanics When `feedback(names, outcome)` is called: - `outcome="delivered"` → effectiveness moves toward 1.0 - `outcome="blocked"` → effectiveness moves toward 0.0 - Uses EMA update: `new_eff = old_eff * 0.9 + outcome * 0.1` - `use_count` increments, `last_used` updates ## Database Location ```bash # Default location .helix/helix.db # Inspect directly sqlite3 .helix/helix.db "SELECT name, effectiveness, use_count FROM insight ORDER BY effectiveness DESC LIMIT 10" ```