verify-bidirectional

What problem does it solve?

When a user refutes or refines an AI recommendation after it has been committed, the AI's persistent baseline (rules, project assets, past decisions) can become stale or inconsistent, leading to conflicting outputs in future sessions. This Skill eliminates that risk by automating the process of updating baselines to reflect user refinements immediately.

Core Features & Use Cases

• Automatic Triggering: Activates naturally when users raise refinement challenges with phrases like "is that right?" or "re-examine this", or via explicit /verify-bidirectional call.
• Evidence-Gated Baseline Updates: Requires cited evidence or cross-asset contradictions before overwriting persistent baseline assets, preventing unsubstantiated changes.
• Human-in-the-Loop Approval: Generates a diff of all proposed baseline changes for user confirmation before applying modifications, ensuring full oversight.
• Use Case: If an AI recommends a project deployment strategy and the user points out a conflict with an existing team security rule, the Skill will surface the conflict, propose updates to the relevant baseline assets, and request approval before saving changes.

Quick Start

When you disagree with an AI recommendation, say "re-examine this" to trigger automatic baseline validation and update proposals.