메뉴
Provide feedback on CYNIC judgments to improve future evaluations. Use when asked to correct a judgment, provide feedback, mark as correct/incorrect, or help CYNIC learn from mistakes.
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
SOC 직업 분류 기준
Transform Claude Code into an AI Scientist that orchestrates research workflows using tree-based hypothesis exploration. Triggers on "research project", "scientific experiment", "run experiments", "AI scientist", "tree search experimentation", "systematic study".
Research methodology before coding: digest → search → patterns → wisdom → ecosystem. For when you need to research before any code, ensuring no ideas from the 500k lines are lost.
Analyze code for simplification: orphans, hotspots, giants, duplicates. 'Don't extract, burn' — three similar lines beat a premature abstraction. Use when asked to simplify, reduce complexity, or clean up code.
Evaluate code, decisions, or content with 36-dimension φ-bounded scoring. 5 axioms, geometric mean Q-Score, max 61.8% confidence. Verdicts: HOWL/WAG/GROWL/BARK. Use when asked to judge, assess, rate, score, or evaluate quality.
Philosophical grounding for technical decisions across 19 traditions. Cross-domain synthesis mapped to 5 axioms (FIDELITY/PHI/VERIFY/CULTURE/BURN). φ-bounded at 61.8%. Use when facing dilemmas, questioning assumptions, or seeking wisdom.
Evaluate any item using CYNIC's 36-dimension judgment system (5 axioms × 7 dims + THE_UNNAMEABLE). Use when asked to judge, evaluate, assess, rate, score, or analyze the quality of code, tokens, decisions, patterns, or any content. Returns Q-Score (0-100), verdict (HOWL/WAG/GROWL/BARK), and dimension breakdown.
| name | learn |
| description | Provide feedback on CYNIC judgments to improve future evaluations. Use when asked to correct a judgment, provide feedback, mark as correct/incorrect, or help CYNIC learn from mistakes. |
| user-invocable | true |
"φ learns from φ"
/learn <judgment_id> correct|incorrect [reason]
Provides feedback to improve CYNIC's judgment accuracy:
| Outcome | When to Use |
|---|---|
correct | Judgment was accurate |
incorrect | Judgment was wrong |
partial | Some parts right, some wrong |
/learn jdg_abc123 correct
/learn jdg_abc123 incorrect "missed security vulnerability"
/learn jdg_abc123 incorrect --score 35 "should have been BARK not WAG"
Use the brain_cynic_feedback MCP tool:
brain_cynic_feedback({
judgmentId: "jdg_abc123",
outcome: "correct|incorrect|partial",
reason: "explanation of feedback",
actualScore: 35 // What it should have been (0-100)
})
CYNIC uses feedback to:
Check learning state:
brain_learning({ action: "state" })
View detected biases:
brain_learning({ action: "biases" })
Trigger calibration:
brain_learning({ action: "calibrate" })
| Feedback Volume | Learning Effect |
|---|---|
| 10+ feedbacks | Initial calibration |
| 50+ feedbacks | Bias detection |
| 100+ feedbacks | Refined accuracy |
/judge - Create judgments to learn from/patterns - See learned patterns/health - Check learning system status