Runs on: Both · Category: quality

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.

After a non-trivial task, the agent rates its own output on 5 axes (accuracy, completeness, clarity, actionability, conciseness) with evidence, producing a 1-5 scorecard plus improvement suggestions. Reflection, not a pass/fail gate.

Cross-harness compatibility

• Hermes: pairs with hermes curator — the self-improvement loop can refine this skill over time.
• OpenClaw: runs via the Node gateway per-agent workspace; no curator, so the skill stays static but functions identically.
• Pure-instruction skill (no language-specific scripts) — runs anywhere both harnesses do.

Verified: Jul 31, 2026 · Hermes 0.16.0 · OpenClaw 2026.7.1-2