EPSS MCP Server
Provides EPSS scores, percentiles, and time series for CVEs, along with vulnerability details from the NVD API, enabling risk assessment through natural language.
README
1. EPSS MCP Server
- 1. EPSS MCP Server
- 2. Features
- 3. Run MCP Server on Docker
- 4. Installation
- 5. Tool Examples
- 6. Reference
Note: The major design of code in this repository is based on EPSS-MCP.
2. Features
EPSS MCP Server provides the following features:
- Retrieve vulnerability details (including description, CWE, and CVSS scores) from the NVD API, and EPSS scores and percentiles from the EPSS API.
- Retrieve EPSS scores for multiple CVEs.
- Retrieve the EPSS time series data for a single CVE.
- Retrieve the top N CVEs with the highest EPSS scores.
3. Run MCP Server on Docker
Note: If you want to change transport type, edit Dockerfile like this.
# Stdio transport(Default)
CMD ["python3", "epss_mcp.py", "--transport", "stdio"]
# SSE transport
CMD ["python3", "epss_mcp.py", "--transport", "sse"]
# Streamable HTTP transport
CMD ["python3", "epss_mcp.py", "--transport", "http"]
Build docker image
cd epss-mcp
docker build -t epss-mcp .
Run MCP server
# Stdio transport
docker run --rm -i epss-mcp
# SSE transport (Access from localhost only is recommended)
docker run -d -p 127.0.0.1:8000:8000 epss-mcp
# Streamable HTTP transport (Access from localhost only is recommended)
docker run -d -p 127.0.0.1:8000:8000 epss-mcp
4. Installation
4.1. Stdio
Prepare a Python virtual environment using uv. See uv for more details.
curl -LsSf https://astral.sh/uv/install.sh | sh
uv sync
. .venv/bin/activate
Cursor, Windsurf
{
"mcpServers": {
"epss-mcp": {
"command": "docker",
"args": [
"run",
"--rm",
"-i",
"epss-mcp"
]
}
}
}
4.2. Streamable HTTP, SSE
Cursor
{
"mcpServers": {
"epss-mcp": {
"url": "http://localhost:8000/mcp"
}
}
}
Windsurf
serverUrl is your server URL or IP address.
{
"mcpServers": {
"epss-mcp": {
"serverUrl": "http://localhost:8000/sse"
}
}
}
Windsurf supports two transport types for MCP servers: stdio and /sse https://docs.windsurf.com/windsurf/cascade/mcp
5. Tool Examples
This example is simply demonstrating how each function works, so it's executed using a Python script as the MCP client. (You can use them via an LLM through editors like Cursor or Windsurf.)
Note: MCP client for testing purposes. (Gemini API Key required)
export GOOGLE_API_KEY=<your api key>
Start the MCP server with stdio transport
$ uv run client/client.py epss-mcp/epss_mcp.py
Connected to server with tools: ['get_cve_info', 'top_epss_cves']
MCP Client Started!
Type your queries or `quit` to exit.
Query:
5.1. Get EPSS Score of single CVE
Query: Get EPSS of CVE-2025-33053
> Thinking...
[Calling tool get_cve_info with args {'cve_id': 'CVE-2025-33053'}]
The EPSS score for CVE-2025-33053 is 0.41763, and the percentile is 97.264.
5.2. Get EPSS Score of multiple CVEs
Query: Get EPSS of CVE-2025-33053, CVE-2025-21298
> Thinking...
[Calling tool get_cve_info with args {'cve_id': 'CVE-2025-33053,CVE-2025-21298'}]
CVE-2025-33053 has an EPSS score of 0.41763 and is in the 97.264 percentile. CVE-2025-21298 has an EPSS score of 0.70558 and is in the 98.593 percentile.
5.3. Get EPSS Score of multiple CVEs with time series
Query: Get time series of CVE-2025-33053 with table format
> Thinking...
[Calling tool get_cve_info with args {'cve_id': 'CVE-2025-33053', 'time_series': True}]
Here is the time series data for CVE-2025-33053 in a table format:
| Date | EPSS | Percentile |
|------------|------------|------------|
| 2025-07-07 | 0.417630000 | 0.972650000 |
| 2025-07-06 | 0.37155000 | 0.96976000 |
| 2025-07-01 | 0.32398000 | 0.96662000 |
| 2025-06-30 | 0.30219000 | 0.96437000 |
| 2025-06-23 | 0.15008000 | 0.94213000 |
| 2025-06-18 | 0.16497000 | 0.94551000 |
| 2025-06-16 | 0.18266000 | 0.94860000 |
| 2025-06-15 | 0.32831000 | 0.96637000 |
| 2025-06-13 | 0.27895000 | 0.96193000 |
| 2025-06-12 | 0.53232000 | 0.97806000 |
| 2025-06-11 | 0.54359000 | 0.97862000 |
5.4. Get top 5 CVEs with the highest EPSS scores
uv run client_stdio.py ../epss-mcp/epss_mcp.py --top_n 5
> Thinking...
[Calling tool top_epss_cves with args {'top_n': 5}]
Here are the top 5 CVEs with the highest EPSS scores:
CVE-2023-42793, EPSS score: 0.94584, EPSS percentile: 100.0
CVE-2024-27198, EPSS score: 0.94577, EPSS percentile: 100.0
CVE-2023-23752, EPSS score: 0.94532, EPSS percentile: 100.0
CVE-2024-27199, EPSS score: 0.94489, EPSS percentile: 99.99900000000001
CVE-2018-7600, EPSS score: 0.94489, EPSS percentile: 99.99900000000001
6. Reference
- https://github.com/jgamblin/EPSS-MCP
- https://nvd.nist.gov/developers/vulnerabilities
- https://www.first.org/epss/api
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.