mcpField
S
stig-mcp

stig-mcp

Local
github.com/jenericSecurity1Updated 3w ago

About stig-mcp

stig-mcp is a local MCP server that answers security questions from a knowledge base built from published guidance. It is intended for security assessors, system owners, and SOC analysts, and maps MITRE ATT&CK to NIST controls, DISA STIG steps, mitigations, and detections without connecting to user systems.

Key features

  1. 01Map ATT&CK techniques to NIST controls
  2. 02Find applicable DISA STIG benchmarks
  3. 03Provide configuration check and fix text
  4. 04Relate mitigations to log-based detections

Use cases

  1. 01Assess controls for an ATT&CK technique
  2. 02Find STIG steps for Windows or RHEL
  3. 03Evaluate detection options from collected logs

MCP support

The local MCP server gives AI assistants access to its security knowledge base. It can be launched by VS Code, Claude Code, or other stdio-compatible MCP clients.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
"mcpServers": {
"jeneric-stig": {
"command": "uvx",
"args": [
"stig-mcp"
],
"env": {
"STIG_MCP_DATA": "<STIG_MCP_DATA>"
}
}
}
}
Run commanduvx stig-mcp
Environment variables to set
STIG_MCP_DATADirectory holding the knowledge base and downloaded sources

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
pypistig-mcp0.2.0

Tools

9

Saved tool list · updated 2w ago

ParameterTypeRequiredDescription
severityanyNo
platformsanyNo
log_sourcesanyNo
technique_idstringYes
benchmark_idsanyNo
system_descriptionanyNo

Details

From the project README

stig-mcp

A local MCP (Model Context Protocol) server that lets an AI assistant, such as GitHub Copilot in VS Code or Claude Code, answer security questions from published guidance instead of from memory. Name a MITRE ATT&CK® technique or threat actor and the system you care about, and the answer gives you:

  • the NIST SP 800-53 security controls that mitigate it;
  • the DISA STIG configuration rules that implement those controls on that system, most severe first, with DISA's check and fix text on request;
  • ATT&CK's own mitigations, and which of its detections work with the logs you collect.

It does not scan or connect to your systems. Its answers come from a knowledge base, a local database of about 6 MB built from those public sources, which you install once. The user guide explains what it does and shows three ways to use it: as a security controls assessor, a system owner, or a SOC analyst.

Prerequisite

stig-mcp is published on PyPI and runs through uvx, which comes with uv. Install uv, then check that uvx --version runs in a new terminal. Restart VS Code or Claude Code after installing uv so it sees the new PATH.

Quick start

VS Code (GitHub Copilot)
  1. Click this badge. VS Code opens and offers to install stig-mcp; choose Install.

    Install in VS Code

  2. Open Copilot Chat and set the mode dropdown to Agent. MCP tools are not available in Ask or Edit mode.

  3. Ask: Install the stig-mcp knowledge base. This is a one-time download of about 6 MB. Allow the tool when VS Code asks.

If that doesn't work, see troubleshooting or setting up VS Code by hand.

Claude Code
  1. Inside a Claude Code session (2.1.275 or later), run:

    /plugin install stig-mcp --marketplace jeneric/STIG-MCP
    
  2. Ask: Install the stig-mcp knowledge base. This is a one-time download of about 6 MB. Allow the tool when Claude Code asks.

If that doesn't work, see troubleshooting or setting up Claude Code by hand.

Example prompts

With the knowledge base installed, try:

  1. System owner: What DISA STIG steps mitigate T1078 on Windows 11?
  2. Security controls assessor: Which ATT&CK techniques does APT29 use?
  3. Anyone, to find the right STIG: Which STIG benchmarks apply to RHEL 9?
  4. SOC analyst: What can I detect of APT29 on Windows Server 2022 with Security and Sysmon logs?

More example prompts are in the user guide, with how to get more out of it and how to make sure the agent answers from the server.

Other MCP clients

Use this only if you are not using VS Code or Claude Code. Any client that launches a local (stdio) MCP server works with this configuration, shown in the common mcpServers shape:

{
  "mcpServers": {
    "stig-mcp": {
      "command": "uvx",
      "args": ["stig-mcp"],
      "env": { "UV_SYSTEM_CERTS": "true", "UV_NATIVE_TLS": "true" }
    }
  }
}

The two environment variables let uv download stig-mcp behind a TLS-inspecting proxy. They are harmless on most other hosts; that section names the one exception. Then ask the agent to install the stig-mcp knowledge base, as in the quick start.

docs/install.md has the details:

What this server fetches

  • The MCP server contacts nothing unless check_sources or install_knowledge_base is called. Then it sends HTTPS GET requests to api.github.com (this repository's release listing) and github.com (/jeneric/STIG-MCP/releases/download/...), which redirects to release-assets.githubusercontent.com or objects.githubusercontent.com. Any other URL, a redirect included, is refused. Nothing is uploaded, and there is no telemetry. install_knowledge_base given a file path and its SHA-256 requests nothing at all.
  • stig-mcp-install-kb contacts the same hosts, and nothing at all with --file.
  • stig-mcp-fetch, used only to build the knowledge base yourself, downloads from raw.githubusercontent.com and api.github.com (MITRE ATT&CK, the CTID mapping, the NIST 800-53 catalog) and from dl.dod.cyber.mil (DISA).

PRIVACY.md states what each of these requests sends and what is stored locally.

Documentation

  • docs/install.md: installation details for every client, where the data lives, and troubleshooting.
  • docs/user-guide.md: for a person talking to an LLM that has this server wired in.
  • docs/operations.md: for whoever installs, builds and maintains the knowledge base.
  • SECURITY.md: reporting a vulnerability, and what is in scope.
  • CONTRIBUTING.md: working from a source checkout, running the tests, and the project's conventions.
  • RELEASING.md: for the maintainer, publishing the package to PyPI and the MCP Registry.
  • PRIVACY.md: what the server and the fetch tool contact, and what is stored locally.

Third-party content

The knowledge base aggregates MITRE ATT&CK, CTID mapping, DISA STIG, DISA CCI list, and NIST OSCAL content. See NOTICE for attribution and licensing obligations and licenses/apache-2.0.txt for the Apache 2.0 license text that notice requires.

Development

Developed with the assistance of Claude Code (Anthropic). All changes were reviewed and tested by the maintainer.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.