Version: v4.0.0
Repository: The-JDdev/manusclaw
License: See repository for details
ManusClaw is a powerful, extensible AI agent framework built in Python that provides a unified interface for interacting with multiple large language model (LLM) providers. It enables you to build, deploy, and manage AI-powered agents that can execute tasks, write code, search the web, manage files, and much more — all from an interactive command-line shell or in headless server mode.
ManusClaw supports a wide range of LLM providers out of the box, including OpenAI, Anthropic, Google, Mistral, AWS Bedrock, Ollama, GGUF, HuggingFace, and Universal/OpenRouter. Whether you want to run models locally through Ollama or connect to cloud-based APIs, ManusClaw has you covered.
- Multi-provider support — Seamlessly switch between OpenAI, Anthropic, Google, Mistral, Bedrock, Ollama, GGUF, HuggingFace, and OpenRouter
- Interactive shell — Full-featured REPL with auto-completion, syntax highlighting, and slash commands
- Persistent task system — Run tasks in the background, manage queues, and resume sessions
- Memory system — Built-in memory management with MEMORY.md and USER.md for persistent context
- Skills system — Extend ManusClaw with custom skills and tools
- Docker support — Run ManusClaw in containers for reproducible, isolated environments
- Search integration — Built-in web search via DuckDuckGo and other providers
- Permission modes — PLAN mode for safe planning, BUILD mode for full execution
- Cron scheduling — Schedule recurring tasks with the built-in cron system
- Multi-agent orchestration — Run multiple agent instances with manusclaw-multi
This repository contains comprehensive setup and usage documentation for ManusClaw. Each document is written to be beginner-friendly while still providing the depth that advanced users need. Every guide includes copy-paste-ready commands, detailed explanations, and troubleshooting tips.
| Document | Description |
|---|---|
| Installation Guide | Step-by-step installation instructions for every platform: Linux, macOS, Windows, Docker, Termux, Colab, and VPS/cloud providers. Includes prerequisite setup, Python configuration, and verification steps. |
| Configuration Guide | Complete reference for configuring ManusClaw. Covers config.toml, .env files, all LLM provider setups, API key management, credential pools, search engines, token budgets, permission modes, and environment variables. |
| Usage Guide | How to use ManusClaw day-to-day. Interactive shell, single-shot mode, all slash commands, background tasks, task queues, persistent tasks, session management, the memory system, skills, and the complete tool reference. |
| Document | Description |
|---|---|
| Deployment Guide | Production deployment walkthrough: Docker, docker-compose, VPS setup, systemd services, Nginx reverse proxy, SSL/TLS, auto-start, process management with supervisord, security hardening, resource planning, and scaling. |
| Document | Description |
|---|---|
| Troubleshooting Guide | Solutions to common problems: installation errors, Python version issues, dependency conflicts, API key problems, Ollama connectivity, Playwright issues, permissions, memory errors, network/firewall, platform-specific quirks, rate limiting, token budgets, and clean reinstall. |
| Uninstall Guide | Complete removal instructions: pip uninstall, config cleanup, workspace removal, Docker image cleanup, and full data purge. |
| Document | Description |
|---|---|
| Termux (Android) Guide | Detailed setup for running ManusClaw on Android devices via Termux, including proot/chroot considerations and performance tips. |
| Google Colab Guide | Step-by-step Colab notebook setup for running ManusClaw in the cloud with free GPU access, including ngrok tunneling for remote access. |
| WSL2 Guide | Windows Subsystem for Linux 2 deep-dive, covering installation, GUI support, file system performance, and integration with Windows tools. |
If you just want to get running as fast as possible, here's the absolute minimum:
# 1. Ensure Python 3.11+ is installed
python3 --version
# 2. Install ManusClaw
pip install manusclaw
# 3. Set your API key (example: OpenAI)
export OPENAI_API_KEY="sk-your-key-here"
# 4. Launch ManusClaw
manusclawFor detailed instructions tailored to your specific platform and use case, follow the Installation Guide.
| Requirement | Minimum | Recommended |
|---|---|---|
| Python | 3.11 | 3.12+ |
| RAM | 512 MB | 2 GB+ |
| Disk Space | 500 MB | 2 GB+ (with Playwright browsers) |
| OS | Linux, macOS, Windows | Linux (Ubuntu 22.04+) |
| Network | Required for API providers | Stable broadband |
Note: If you plan to run local models via Ollama, you will need significantly more RAM and ideally a GPU. See the Configuration Guide for Ollama setup details.
When you install and run ManusClaw, the following directory structure is created:
~/.manusclaw/ # Configuration directory
├── config.toml # Main configuration file
├── .env # Environment variables (API keys)
├── MEMORY.md # Agent persistent memory
├── USER.md # User profile and preferences
└── skills/ # Custom skills directory
workspace/ # Default working directory
├── ... # Your project files
Found an error or want to improve these docs? Contributions are welcome! Please open an issue or pull request at the main repository.
- GitHub Issues: The-JDdev/manusclaw/issues
- Documentation: You're looking at it! Start with the Installation Guide.
- Troubleshooting: Check the Troubleshooting Guide before opening an issue.