MedgeClaw is an open-source AI research assistant designed for biomedicine, enabling users to converse and automate complex analyses such as RNA-seq, drug discovery, and clinical research. It targets researchers and power users, offering a significant benefit by integrating a broad set of scientific skills with interactive development environments and real-time progress tracking.

How It Works

The system employs a layered architecture: OpenClaw acts as the conversational gateway, connecting to messaging apps like WhatsApp or Slack. Claude Code serves as the execution layer, autonomously running workflows powered by 140 K-Dense Scientific Skills. Analyses are performed within a Dockerized R and Python environment, providing access to tools like DESeq2, Seurat, Scanpy, and scikit-learn. This approach offers a novel way to interact with and control sophisticated bioinformatics pipelines through natural language, delivering results directly into RStudio or JupyterLab, complemented by a real-time dashboard.

Quick Start & Requirements

• Primary install/run command: Clone with submodules (git clone --recurse-submodules), run bash setup.sh twice (filling in API key in .env after the first run), then start the environment with docker compose up -d and the assistant with openclaw onboard.
• Non-default prerequisites: Node.js 22+, Docker + docker-compose, Git, and an API key from a supported model provider (Anthropic Claude, MiniMax, GLM-4.7, DeepSeek, or Ollama for local execution).
• Links: Node.js (nodejs.org), Docker (docs.docker.com).

Highlighted Details

• Features 140 K-Dense Scientific Skills covering genomics, drug discovery, clinical research, and machine learning.
• Provides a real-time Research Dashboard for live progress tracking, code preview, and output viewing.
• Integrates directly with RStudio Server (localhost:8787) and JupyterLab (localhost:8888) for interactive analysis.
• Supports conversational interaction via WhatsApp, Slack, Feishu, and Discord.
• Includes CJK visualization for matplotlib and SVG UI templates for professional reporting.

Maintenance & Community

The project includes a detailed roadmap with many features already implemented, such as the core architecture, Docker environment, research dashboard, and third-party API proxy support. Future plans include multi-agent workflows, automated literature integration, and interactive report builders. No direct community links (e.g., Discord, Slack) are provided in the README.

Licensing & Compatibility

The project is licensed under MIT. It bundles K-Dense Scientific Skills as a git submodule, which is also MIT licensed, but individual skills within that submodule may have their own licenses, requiring users to check each SKILL.md file for details. The MIT license generally permits commercial use and closed-source linking.

Limitations & Caveats

When using third-party API proxies (MiniMax, GLM, DeepSeek), users must configure ANTHROPIC_SMALL_FAST_MODEL in the .env file to prevent silent failures in the BashTool pre-flight check. Users are also advised to verify the licenses of individual skills within the scientific-skills submodule.