Reflect Check

What it does

The reflect-check skill allows an AI agent to verify its own reasoning and identify potential errors by prompting itself with a follow-up question based on its previous response. This process helps the agent detect inconsistencies or gaps in logic, leading to more reliable outputs. It's designed to improve the quality of responses through self-assessment.

When to use it

• Complex Reasoning Tasks: Use when an AI needs to perform multi-step reasoning or analysis where errors are likely.
• Creative Writing: Employ to ensure narrative consistency and plot coherence in generated stories or scripts.
• Code Generation: Apply to check for logical flaws or potential bugs in the code produced by the agent.
• Fact Verification: Utilize when accuracy is paramount, such as summarizing research papers or answering factual questions.

Key capabilities

• Self-prompting: Generates follow-up prompts based on previous responses.
• Error detection: Identifies inconsistencies and potential errors in reasoning.
• Improved reliability: Enhances the overall quality and trustworthiness of AI outputs.

Example prompts

• "Analyze this document and then reflect-check your analysis."
• "Write a short story, and after each paragraph, reflect-check for consistency."
• "Generate Python code to solve this problem; reflect-check the logic before presenting it."

Tips & gotchas

The effectiveness of reflect-check depends on the agent's ability to understand its own reasoning process. It may require some experimentation with prompt phrasing to achieve optimal results.