mmi

mmi (Mother May I?) is a CLI utility that reduces friction when using Claude Code by automatically approving known-safe shell commands while maintaining security through a configurable allowlist and deny-list system. It uses AST-based parsing to correctly handle pipes, command chains, and shell constructs, ensuring each segment is validated independently.

Designed for developers tired of manually approving repetitive commands like git status, pytest, and ls, mmi implements a three-layer model: deny patterns are always rejected, safe commands are auto-approved, and everything else prompts for approval. The maker built it with Claude Code and notes that while it reduces risk, it works best alongside Claude Code's built-in sandbox mode for additional control.