XAgent is an open-source autonomous LLM agent designed for complex task solving, targeting users who need automated assistance with multi-step processes. It aims to provide a general-purpose agent capable of handling diverse tasks, from data analysis and coding to acting as a life assistant, with a focus on autonomy, safety, and extensibility.

How It Works

XAgent employs a modular architecture comprising a Dispatcher, Planner, and Actor. The Dispatcher manages task instantiation and routing to different agents. The Planner breaks down complex tasks into subtasks and milestones, while the Actor executes these subtasks using a suite of tools provided by ToolServer. This approach allows for step-by-step problem-solving and dynamic adaptation.

Quick Start & Requirements

• Installation: Requires Python >= 3.10. Install dependencies via pip install -r requirements.txt.
• ToolServer: Must be set up using Docker and docker-compose. Pull the image with docker compose up or build locally with docker compose build followed by docker compose up.
• Configuration: Requires an OpenAI API key (GPT-4 or GPT-4-32k recommended) configured in assets/config.yml.
• Execution: Run via python run.py --task "your task here" --config-file "assets/config.yml".
• GUI: Accessible at http://localhost:5173 after starting the XAgent-Server container (default credentials: guest/xagent).
• Documentation: Tutorial, Demo, Documentation.

Highlighted Details

• ToolServer: Provides a sandboxed Docker environment for executing actions, including file editing, Python notebooks, web browsing, shell commands, and RapidAPI integrations.
• Human Collaboration: Features a "cooperation with human" mode where the agent can request assistance or guidance.
• Evaluation: Claims superior performance over AutoGPT in human preference evaluations across categories like Data Analysis, Coding, and Math.
• Record Keeping: Logs all intermediate steps, LLM interactions, and tool usage for reproducibility and sharing.

Maintenance & Community

• Community: Active Discord server (https://discord.gg/zncs5aQkWZ) and welcome contributions.
• Contributors: Acknowledges a list of contributors.
• Roadmap: Implied through ongoing development and feature additions.

Licensing & Compatibility

• License: Apache 2.0.
• Compatibility: Permissive license suitable for commercial use and integration with closed-source projects.

Limitations & Caveats

The project is described as being in its "early stages," suggesting potential for instability or missing features. Recommended LLM models are proprietary (OpenAI), and reliance on specific versions of these models may impact long-term usability.