What problem does it solve? AI coding agents sometimes run commands or take actions you would rather block, gate, or augment with context. This Skill turns those concerns into concrete hook definitions stored in .github/hooks/ so policies are enforced automatically instead of repeated in every conversation. ## Core Features & Use Cases - Policy Extraction from Conversation: Reviews the current conversation for concerns like "don't run this command" or "always check before X" and generalizes them into a hook. - Lifecycle Event Automation: Creates hooks for events such as PreToolUse, SessionStart, or Stop that can block, warn, or inject context. - Iterative Refinement: Drafts the hook JSON and any companion scripts, asks about ambiguous parts, then summarizes what the hook enforces and how to test it. - Use Case: After telling an agent twice not to run destructive git commands, invoke this Skill to produce a PreToolUse hook in .github/hooks/ that blocks those commands permanently. ## Quick Start Ask the agent to create a hook that blocks force pushes to the main branch before any git command runs.