| name | meta-prompting |
| description | Agents optimizing prompts for other agents (Chain of Thought, Tree of Thoughts). |
| role | orch-planner |
| triggers | ["optimize prompt","better prompt","chain of thought","system prompt","prompt engineering"] |
meta-prompting Skill
This skill enables "Meta-Cognition": thinking about how to think.
1. Optimization Strategies
- CoT (Chain of Thought): "Let's think step by step." (Force reasoning before answer).
- ToT (Tree of Thoughts): "Generate 3 possible solutions. Evaluate each. Pick the best."
- Few-Shot: Provide 3 examples of "Input -> Output" to guide the model.
2. Dynamic Prompt Construction
Instead of static prompts, build them based on context:
- Role: "You are a world-class [Role]."
- Context: "We are building [Project]. The tech stack is [Stack]."
- Constraint: "Do not use [Forbidden Lib]. Answer in [Format]."
3. Self-Correction Prompts
- "Review your previous answer. Did you meet all requirements? If not, rewrite it."
- "Are there any security vulnerabilities in the code you just wrote?"
4. Tool Use Prompts
- "You have access to
grep. Use it to find the definition of User class before guessing."