claude-trust-me-bro

"Can I run this command?" — "Trust me bro."

A CLI tool that automatically approves all Claude Code permission prompts. One command to skip every confirmation dialog and let Claude Code run fully autonomous.

Tired of clicking "yes" on every file read, shell command, and tool call? Just tmb enable and get back to work.

The Problem

Claude Code is amazing, but it asks for permission constantly:

• "Can I read this file?" Yes.
• "Can I run this shell command?" Yes.
• "Can I write to this file?" Yes.
• "Do you want to proceed?" YES.

When you're doing trusted work — supervised coding sessions, batch refactors, CI/CD pipelines, or long autonomous tasks — these prompts kill your flow.

The Solution

npx claude-trust-me-bro enable

That's it. No more prompts. Claude Code just... does its thing.

How It Works

The tool uses a three-layer approach to make sure no permission prompt slips through:

| Layer | Mechanism | What It Does | |-------|-----------|--------------| | 1 | PreToolUse hook | Auto-approves tool calls before the permission check fires | | 2 | PermissionRequest hook | Auto-approves "Do you want to proceed?" dialogs | | 3 | Allow rules in settings.local.json | Broad wildcard permissions for all built-in and MCP tools |

All three layers are installed and removed together — no partial states, no leftovers.

Installation

# Install globally
npm install -g claude-trust-me-bro

# Or just use npx (no install needed)
npx claude-trust-me-bro enable

Usage

The CLI command is tmb — short, sweet, and to the point.

# Enable auto-approve — trust me bro
tmb enable

# Check if Claude trusts you
tmb status

# Revoke trust — bring back permission prompts
tmb disable

After enabling or disabling, restart Claude Code for changes to take effect.

What Gets Auto-Approved

Every built-in Claude Code tool and all MCP server tools:

| Tool | Description | |------|-------------| | Bash(*) | Shell commands | | Read(*) | File reading | | Write(*) | File creation | | Edit(*) | File editing | | Glob(*) | File pattern matching | | Grep(*) | Content search | | WebFetch(*) | HTTP requests | | WebSearch(*) | Web searches | | NotebookEdit(*) | Jupyter notebook edits | | mcp__* | All MCP server tools |

Where Changes Are Made

| File | What's Added | |------|-------------| | ~/.claude/settings.json | PreToolUse and PermissionRequest hooks | | ~/.claude/settings.local.json | Wildcard allow rules for tool permissions |

Running tmb disable cleanly removes everything this tool added, without touching your other Claude Code settings.

Frequently Asked Questions

Is this safe?

This tool is designed for supervised, trusted environments. It removes the safety net of permission prompts, so only use it when you're actively watching what Claude Code does, or in controlled environments like CI pipelines.

Does it work with MCP servers?

Yes. The mcp__* wildcard rule covers all MCP server tools automatically, no matter how many servers you have configured.

Can I undo it?

Absolutely. Run tmb disable and restart Claude Code. Everything goes back to normal — permission prompts return and all hooks are removed.

Does it survive Claude Code updates?

The hooks and rules are stored in your user-level Claude Code settings, which persist across updates.

Requirements

• Node.js >= 18
• Claude Code CLI installed

Related

• Claude Code Documentation — Official docs for Claude Code
• Claude Code Hooks — How hooks work in Claude Code

License

MIT