agent-self-evaluation
Use after completing any non-trivial task. The agent self-rates its output on 5 axes — accuracy, completeness, clarity, actionability, conciseness — with concrete evidence per criterion. Produces a structured 1-5 scorecard with specific improvement suggestions.
- installs 8w
- 1,727
- 30-day movement
- starts with the next reading
- Related entries
- 1
- Connections
- 0
A SKILL.md that makes an AI coding agent pause after non-trivial tasks and self-rate its output on five axes: accuracy, completeness, clarity, actionability, and conciseness. It enforces an evidence rule — any score below 5 must cite concrete evidence — and produces a structured 1-5 scorecard with improvement suggestions.
Reach for it when you want the agent's self-assessment to be evidence-based rather than reflexive all-5s, since the skill explicitly defines anti-patterns like unevidenced perfect scores.
Use it to
- Score a multi-file code change against the 5-axis rubric
- Generate an evaluation report after a multi-step implement-test-review workflow
- Apply one concrete improvement per axis scoring 3 or below
- Audit debugging-session output for completeness and correctness gaps
For Developers configuring AI coding agents for self-review
- Host repository
- affaan-m/ECC
- Installs, lifetime
- 2,200
- Installs, 8 weeks
- 1,727
- Host stars
- 259k
- Host language
- JavaScript