EU_AI_ACT_MCP

This MCP server provides automated EU AI Act compliance tools that help you classify AI systems by risk, determine your role under the Act, verify prohibited practices, and add transparency disclosures. It also includes content watermarking, deepfake labeling, and security checks to detect prompt injections and sensitive file access. Tools are organized into categories such as Risk & Role Classification, Transparency & Disclosure, Content Watermarking, Deepfake Labeling, and AI Security, making it straightforward to call the exact function you need for governance and transparency tasks. You can run this server locally, host it yourself, or integrate it into your AI workflows to streamline compliance workflows.

Prerequisites:

• Python 3.10+ installed on your system
• A virtual environment tool (optional but recommended)
• Internet access to install dependencies

1. Create and activate a virtual environment (recommended):

python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

2. Install dependencies from requirements.txt:

pip install -r requirements.txt

3. (Optional) Run tests or initialize any required data as described in the project docs:

python test_all_tools.py

4. Start the MCP server:

python server.py

5. Connect clients (e.g., Claude Desktop, Windsurf) using the example configurations found in the README, pointing to the running server and providing the appropriate environment variables if needed.

Environment variables like SONNYLABS_API_TOKEN and SONNYLABS_ANALYSIS_ID are used for optional SonnyLabs security tooling integration. If you don’t need SonnyLabs features, you can omit them or replace with your own credentials as appropriate. Common issues include port conflicts, missing dependencies, or firewall restrictions preventing external tools from accessing the server. Ensure your Python version, virtual environment, and working directory are correctly set. For production deployments, consider containerization (Docker) or orchestrated deployment, and secure your API tokens using secret management best practices.