sycophancy-guard

Sycophancy in AI conversations manifests as thesis inflation, echo chambers, and unverified assumption stacking. This skill detects these patterns to prevent models from mirroring user biases and escalating agreement without evidence, which leads to delusional decision-making.

You audit AI assumptions in market analysis by maintaining a structured log of strategic positions and their evidence basis. This skill integrates with memory stores to log belief audits, ensuring long-term consistency and accountability for high-risk claims.

Yes, you can trigger an adversarial review automatically when high-risk strategic claims are made. This skill automatically initiates a Decision Council or forces a confidence declaration to intervene during critical geopolitical forecasting or market analysis.

Yes, bias detection for AI belief spiraling requires integration with the system's memory store and decision-making protocols. This connection is necessary to log belief audits and trigger adversarial reviews effectively across high-risk domains.

You should not rely on automated sycophancy detection when lacking the ability to integrate with system memory stores. Without this connection, the skill cannot log belief audits or trigger adversarial reviews to prevent delusional decision-making.