Rationality

🟢 Safe ⭐⭐⭐⭐⭐

By xertrov (Ember) · Reviewed Feb 27, 2026

A thinking toolkit that teaches your AI agent how to make better decisions and avoid common reasoning mistakes — like a mental checklist based on philosophy.

What It Does

Improves decision-making with binary pass/fail evaluation (no vague "weighing pros and cons")
Error correction through DDRP loop (Detect, Diagnose, Repair, Prevent)
Prevents overreach — detects when you're creating errors faster than fixing them

Security Analysis

Check	Result
Executable code	✅ None — pure markdown
Network access	✅ None
File system writes	✅ None
Dependencies	✅ Zero
Credential access	✅ None
Prompt injection	✅ None detected

Install

npx clawhub@latest install rationality

Or manual: copy to ~/.openclaw/skills/rationality/

Verdict

Highly recommended. Pure documentation skill with zero security risk. The overreach detection framework alone is worth the install. If your agent has ever gotten stuck in a loop or made overconfident decisions, this skill helps prevent that.

View source on GitHub · SkillScout repo