self-eval-bias

Detect self-evaluation bias in agent code reviews by identifying overconfidence patterns and prompting a fresh perspective with concrete evidence. This skill analyzes review outputs to ensure agents do not overvalue their own work due to context bias.

Multi-agent review processes produce overly confident assessments due to context bias from repetitive evaluation scenarios. Agents become overly confident about their own work, requiring adversarial probing and fresh context to ensure a fair, thorough review.

Prevent overconfidence in automated code reviews by applying adversarial probing to outputs and demanding concrete evidence. This approach ensures a fresh perspective is applied, mitigating biased assessments in multi-agent systems.

Ensure a thorough agent review process for code patches by requiring fresh context and concrete evidence. This skill advises on avoiding biased assessments by prompting reviewers to conduct fresh evaluations when overconfidence is detected.

Yes, you need a fresh context to avoid bias in repetitive evaluation scenarios. This skill requires fresh context and concrete evidence to successfully identify and mitigate biased assessments in multi-agent systems.

You should not rely on an agent's own code review for quality assurance when overconfidence patterns emerge from context bias. Adversarial probing of outputs and a fresh perspective are required to ensure a fair and thorough evaluation.