argus-docs-mcp 0.1.0

MCP server for managing Argus project documentation from any AI client

Argus Docs MCP Server

A Model Context Protocol server that connects AI clients to your Argus project documentation. Read, create, update, delete, and search docs directly from Claude, Cursor, Windsurf, or any MCP-compatible client.

Features

• Full CRUD — create, read, update, and delete documentation pages
• Full-text search — search across titles and document content
• Project scoping — auto-scope to a default project or switch between projects
• Category filtering — organize docs as documentation, api_reference, guide, changelog, or note
• Document hierarchy — nest pages under parent documents
• Markdown in/out — write and read docs in markdown (auto-converted to rich text internally)

Prerequisites

1. An Argus instance (self-hosted or cloud)
2. An API key with documents scope — create one in Argus → Settings → API Keys
3. Python 3.10+ or uv/uvx

Quick Start

Install

# With pip
pip install argus-docs-mcp

# Or with uv (recommended)
uv pip install argus-docs-mcp

Configure Your AI Client

Claude Code

Add to your project's .mcp.json:

{
  "mcpServers": {
    "argus-docs": {
      "command": "argus-docs-mcp",
      "env": {
        "ARGUS_API_URL": "https://your-instance.com/api",
        "ARGUS_API_KEY": "argus_your_key_here",
        "ARGUS_PROJECT_ID": "your-project-slug"
      }
    }
  }
}

Or with uvx (no pre-install needed):

{
  "mcpServers": {
    "argus-docs": {
      "command": "uvx",
      "args": ["argus-docs-mcp"],
      "env": {
        "ARGUS_API_URL": "https://your-instance.com/api",
        "ARGUS_API_KEY": "argus_your_key_here",
        "ARGUS_PROJECT_ID": "your-project-slug"
      }
    }
  }
}

Claude Desktop

Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json on macOS, %APPDATA%\Claude\claude_desktop_config.json on Windows):

{
  "mcpServers": {
    "argus-docs": {
      "command": "argus-docs-mcp",
      "env": {
        "ARGUS_API_URL": "https://your-instance.com/api",
        "ARGUS_API_KEY": "argus_your_key_here",
        "ARGUS_PROJECT_ID": "your-project-slug"
      }
    }
  }
}

Cursor / Windsurf

Add the same configuration to your editor's MCP settings file. Check your editor's documentation for the config file location.

Environment Variables

Variable	Required	Description
ARGUS_API_URL	Yes	Base URL of your Argus API (e.g. https://your-instance.com/api)
ARGUS_API_KEY	Yes	API key with documents scope
ARGUS_PROJECT_ID	No	Default project UUID, slug, or name. When set, tools auto-scope to this project.

Available Tools

Tool	Description
list_projects	List all accessible projects with their IDs
list_docs	List documentation pages for a project, filtered by category
read_doc	Read a document's full content as markdown
create_doc	Create a new documentation page from markdown
update_doc	Update an existing page's title and/or content
delete_doc	Delete a documentation page permanently
search_docs	Search docs by title and content with snippets

Usage Examples

Once connected, you can ask your AI client things like:

• "List all the documentation pages in my project"
• "Create a new API reference doc for the /users endpoint"
• "Update the Getting Started guide with the new installation steps"
• "Search the docs for anything about authentication"
• "Delete the outdated deployment guide"

Development

To run locally against a dev Argus instance:

# Clone and install in editable mode
git clone <repo-url>
cd mcp-server
uv pip install -e .

# Set environment variables
export ARGUS_API_URL=http://localhost:5050/api
export ARGUS_API_KEY=argus_your_dev_key
export ARGUS_PROJECT_ID=your-project

# Run
argus-docs-mcp

Troubleshooting

"ARGUS_API_KEY not set" — Add the ARGUS_API_KEY environment variable to your MCP client config.

"API key rejected (401)" — Check that the key hasn't been revoked. Create a new one in Argus → Settings → API Keys.

"API key lacks 'documents' scope (403)" — The key needs the documents scope. Create a new key with the correct scope.

"Cannot reach API" — Verify ARGUS_API_URL points to your Argus instance and the server is running.