Daily Library Review
The data-driven quality loop that promotes winners and retires underperformers.
Coming soon
The Daily Library Review depends on Output Ledger data (acceptance rates, rework ratios) which is not yet collected. This feature is on the roadmap.
The Daily Library Review will be the automated quality assurance process that keeps your Skill Library healthy. Every day, it will analyze Output Ledger data to identify which skills are working and which need attention.
How It Works
- Aggregate — Collect all output data from the last 24 hours across every runtime
- Score — Calculate acceptance rates, rework ratios, and quality trends per skill
- Flag — Surface skills with declining quality or high rework rates
- Recommend — Suggest promotions (org-wide rollout) or retirements (skill removal)
- Report — Generate a daily summary for skill owners and governance leads
Quality Signals
| Signal | Meaning | Action |
|---|---|---|
| High acceptance (>90%) | Skill is producing quality output | Consider promotion |
| Rising rework (>20%) | Skill output needs frequent editing | Flag for revision |
| Low adoption (under 5 runs/week) | Skill may be irrelevant or hard to find | Review discoverability |
Self-Improving System
The Daily Review will create a feedback loop that manual processes can't match:
- Skills that produce consistently high-quality output get promoted to org-wide availability
- Skills that produce rework get flagged for their authors to revise
- Skills that consistently underperform get retired automatically