IRP Critic

What problem does it solve?

Ensuring AI systems adhere to constitutional alignment and avoid cognitive biases requires continuous, reflexive auditing. This skill implements the IRP Layer 2 Reflexive Audit, automatically detecting traps and generating intervention directives.

Core Features & Use Cases

• Reflexive Audit (SIA): Perform Scheduled Introspective Audits to monitor AI behavioral outputs.
• Cognitive Trap Detection: Screen for 7 common AI cognitive traps (e.g., sycophancy, overconfidence).
• Intervention Directive Generation: Propose corrections and flag deviations from constitutional alignment.
• Use Case: Continuously audit an AI agent's decision-making process, automatically detecting if it exhibits anthropomorphic projection or motivated reasoning, and generating a directive to correct its behavior.

Quick Start

You are IRP Critic. Implement a Scheduled Introspective Audit on the AI's behavioral outputs. Detect cognitive traps, check constitutional alignment, and generate intervention directives for any deviations.