mcp_nuclei_server

The MCP Nuclei Server exposes Nuclei-based security scanning as an MCP service. It allows large language models or other MCP clients to trigger Nuclei scans through a standardized MCP interface, with configurable templates, template tags, and severity-based filtering, and returns results in a structured JSON format. You can customize the scan by selecting the templates to apply, optional tag filters, and a severity threshold, then request results for a given target.

To use the service, connect to the MCP endpoint that hosts the nuclei_mcp_server, and issue a request containing the target, desired templates, severity level, optional template_tags, and an output_format (defaulting to JSON). The server will invoke Nuclei with the specified options, collect the findings, and return a JSON payload containing success status, the target, time cost, and an array of results with template names, severities, and details for each finding.

Prerequisites

• Python 3.8 or higher
• Nuclei binary installed and accessible in your PATH or at a known location
• Internet access to install UV (the MCP runtime) and Python dependencies

Installation steps

1. Clone the repository and navigate into it

git clone https://github.com/crazyMarky/mcp_nuclei_server.git
cd mcp_nuclei_server

2. Install UV and set up the Python environment

# Install UV (instructions vary by OS)
curl -LsSf https://astral.sh/uv/install.sh | sh  # Linux/Mac
# Or for Windows use the PowerShell installation as described in the original guide

3. Create and activate a virtual environment, then install MCP-related packages

uv venv
source .venv/bin/activate
# Install MCP related packages
uv pip install mcp

4. Ensure Nuclei is installed and accessible, and configure the mcp_server as needed (see mcp_config example). If you store the nuclei binary in a custom path, export NUCLEI_BIN_PATH accordingly.

# Example: verify nuclei is in PATH
nuclei --version

5. Run or deploy the MCP Nuclei server via your MCP runtime (as configured in mcp_config).

Environment variables and paths must be adjusted to your environment. Common issues include: missing Nuclei binary or incorrect NUCLEI_BIN_PATH, UV not installed or not on PATH, and mismatched Python/UV environments. The MCP configuration uses uvx to run a Python-based server directory; ensure the directory contains the expected run/main.py entry point and that the templates directory and nuclei binary are accessible. If you modify the templates or tags, reload or restart the MCP service to apply changes. When troubleshooting, check the logs emitted by the UV runtime for hints about path errors or missing dependencies.