What problem does it solve? Repeated permission prompts interrupt your workflow and slow down AI-assisted sessions. This Skill analyzes which tools and commands you approve or deny most often and generates targeted permission rules so safe operations run without prompts while dangerous ones stay blocked. ## Core Features & Use Cases - Pattern Analysis: Scans current permission settings in .claude/settings.json and ~/.claude/settings.json to identify frequently approved tools. - Risk-Tiered Rule Generation: Groups rules into safe allow-list candidates (Read, Glob, Grep, read-only git commands), medium-risk ask candidates (Edit, Write, git commit), and high-risk deny candidates (rm -rf, git push --force, git reset --hard). - Approval-First Workflow: Presents all proposed rules for user review before applying them, with an estimate of prompts saved per session. - Use Case: After a long coding session where you manually approved dozens of git status and npm test prompts, run this Skill to auto-approve those read-only and test commands while keeping destructive operations blocked. ## Quick Start Ask the assistant to analyze my recent permission approvals and generate optimized allow and deny rules for my Claude settings.
