Confidence Honesty

What it does

This skill encourages AI agents to respond with a measured level of confidence and prioritize honesty in their answers. It aims to reduce overly assertive or misleading statements by prompting the agent to acknowledge uncertainty when appropriate, and to avoid fabricating information. The skill promotes more reliable and trustworthy responses from the AI.

When to use it

• When you need an AI agent to provide factual information where accuracy is critical (e.g., medical advice, legal guidance).
• To prevent an AI from confidently stating incorrect or unsubstantiated claims in creative writing scenarios.
• In situations requiring nuanced opinions or assessments where acknowledging limitations is important.
• When you want the AI to avoid making up details or "hallucinating" information.

Key capabilities

• Promotes honest responses.
• Encourages measured confidence levels.
• Reduces fabrication of information.
• Acknowledges uncertainty when appropriate.

Example prompts

• “Explain [topic] and be sure to indicate if there's any uncertainty or conflicting information.”
• "Summarize this article, but avoid making definitive statements unless you are certain."
• “Describe the potential risks of [action], and please state your confidence level in each assessment.”

Tips & gotchas

The skill’s effectiveness depends on the underlying AI model's ability to understand and apply nuanced instructions. It is not a guaranteed solution for preventing all inaccuracies, but it can help steer the agent towards more responsible responses.