claude-composer
Claude Code CLI enhancer — automates prompt confirmations, manages toolsets, and configures trusted roots
Claude Composer is a CLI utility for Claude Code developers that streamlines interaction by automating prompt confirmations and managing tool configurations. It allows defining custom toolsets and trusted directories, reducing repetitive manual inputs during development workflows. The tool also provides options for notification control and detailed logging, enhancing the overall Claude Code experience.
- Automates prompt acceptance with a 'YOLO' mode
- Configures custom toolsets for Claude Code
- Defines trusted root directories to bypass prompts
- Provides system notifications for Claude Code events
- Offers detailed logging for debugging and insights
README
View on GitHub ↗Claude Composer
THIS REPO HAS GRADUATED. THAT MEANS THAT I WILL NO LONGER BE USING IT AND WISH ANYONE LUCK AND JOY IF THEY WANT TO FORK IT, CONTINUE USING IT, TAKE IT OVER, OR WHATEVER OTHER ACTIVITIES MAKE A PERSON HAPPY. GOING FORWARD I AM USING HOOKS TO ACHIEVE YOLO MODE:
https://github.com/possibilities/claude-code-generic-hooks
AND I USE PERMISSION MODES TO START IN PLAN MODE:
https://docs.anthropic.com/en/docs/claude-code/iam#permission-modes
ALSO USE DIRTY-CONFIRM TO GET THE SAME CONFIRMATIONS THAT CLAUDE-COMPOSER PROVIDES AROUND NON-GIT AND DIRTY GIT DIRECTORIES:
https://github.com/possibilities/dirty-confirm
A tool that adds small enhancements to Claude Code
Features
- Reduced interruptions: Automatic confirmation prompt acceptance
- Tool management: Toolsets configure which tools Claude can use
- Enhanced visibility: System notifications keep you informed (BROKEN, WILL GET BIG UPDATE SOON)
Quick Start
# Install
npm install -g claude-composer
# Initialize configuration
claude-composer cc-init
# Run with default settings
claude-composer
# Use YOLO mode (accept all prompts)
claude-composer --yolo
Installation
Prerequisites: Node.js 18+, npm/yarn/pnpm, Claude Code installed
# Global install
pnpm add -g claude-composer
# or
yarn global add claude-composer
# or
npm install -g claude-composer
Command Line Options
Core Options
# Configuration
--yolo # Accept all prompts automatically
--toolset <name...> # Use specified toolsets
--mode <mode> # Start in 'act' or 'plan' mode
--ignore-global-config # Ignore global config
# Safety
--dangerously-allow-in-dirty-directory
--dangerously-allow-without-version-control
--dangerously-suppress-yolo-confirmation
# Notifications
--show-notifications / --no-show-notifications
--sticky-notifications / --no-sticky-notifications
# Debug
--quiet # Suppress preflight messages
--allow-buffer-snapshots # Enable Ctrl+Shift+S snapshots
--log-all-pattern-matches # Log to ~/.claude-composer/logs/
All unrecognized options pass through to Claude Code.
See docs/cli-reference.md for complete reference.
Configuration
Run claude-composer cc-init to create configuration:
# Global config (default)
claude-composer cc-init
# Project-specific config
claude-composer cc-init --project
Basic Configuration
# config.yaml
yolo: false # Set to true to accept all prompts
toolsets:
- internal:core
- my-tools
roots:
- ~/projects/work
- ~/projects/personal
show_notifications: true
sticky_notifications: false
See docs/configuration.md for details.
YOLO Mode
When enabled with --yolo flag or yolo: true in config, Claude Composer will automatically accept all prompts without confirmation. Use with caution!
Toolsets
Configure which tools Claude can use and MCP server connections.
See docs/toolsets.md for details.
Trusted roots
Use trusted roots to define trusted parent directories to auto-accept initial trust prompts.
See docs/roots-config.md for details.
Environment Variables
See docs/environment-variables.md for details.
License
This project is in the public domain. See the UNLICENSE file for details.
Similar other
MiroFish
Swarm intelligence engine for future prediction — simulates multi-agent interactions to deduce outcomes
paperclip
AI agent orchestration server for autonomous companies — manages teams of bots, goals, budgets, and governance
learn-claude-code
Agent harness engineering tutorial — build the environment and tools for real LLM agents, inspired by Claude Code
LightRAG
Lightweight RAG framework — builds simple, fast, and scalable retrieval-augmented generation systems for LLMs