SnirRadomsky/confluence-mcp

Confluence MCP Server

An MCP (Model Context Protocol) server for Atlassian Confluence that allows AI assistants like Claude and Cursor AI to interact with your Confluence instance. The server enables querying spaces, retrieving page content, and searching your Confluence knowledge base directly through your AI assistant.

Features

• List available Confluence spaces
• Get detailed information about specific spaces
• Retrieve page content by ID or by title within a space
• Search Confluence using CQL (Confluence Query Language)
• Security-focused design with local credential management

Getting Started

Prerequisites

• Node.js 18 or higher
• Docker (recommended for containerized usage)
• An Atlassian Confluence account with API access
• An Atlassian API token

Installation

1. Clone the repository:

   git clone https://github.com/SnirRadomsky/confluence-mcp.git
   cd confluence-mcp
   

2. Install dependencies:

   npm install
   

   Note: The package-lock.json file ensures consistent dependency versions across environments. It's recommended to keep it in the repository for reproducible builds.

Configuration

1. Copy the env.example file to env:

   cp env.example env
   

2. Edit the env file with your Confluence credentials:

   CONFLUENCE_URL=https://your-organization.atlassian.net/wiki
   CONFLUENCE_USERNAME=your-email@example.com
   CONFLUENCE_API_TOKEN=your-api-token-here
   

   To get an API token:

   • Go to https://id.atlassian.com/manage-profile/security/api-tokens
   • Click "Create API token"
   • Copy the generated token

3. (Optional) If you want to filter available spaces, uncomment and set CONFLUENCE_SPACES_FILTER:

   CONFLUENCE_SPACES_FILTER=DEV,TEAM,DOCS
   

Running the Server

Using Docker (Recommended)

1. Make sure the scripts are executable:

   chmod +x docker-mcp-launcher.sh restart-mcp.sh docker-entrypoint.sh
   

2. Start the server:

   ./docker-mcp-launcher.sh
   

3. To restart the server if needed:

   ./restart-mcp.sh
   

Without Docker

1. Install dependencies:

   npm install
   

2. Start the server:

   npm start
   

Connecting to Your AI Assistant

Update your AI assistant configuration to use this MCP server. For Cursor, edit ~/.cursor/mcp.json:

{
  "mcpServers": {
    "confluence-mcp": {
      "type": "stdio",
      "command": "/path/to/confluence-mcp/docker-mcp-launcher.sh",
      "restartOptions": {
        "command": "/path/to/confluence-mcp/restart-mcp.sh",
        "delay": 1000,
        "maxRestarts": 5
      }
    }
  }
}

Available Tools

The server provides the following MCP tools:

• confluence_get_spaces: List available Confluence spaces
• confluence_get_space: Get detailed information about a specific space
• confluence_get_page: Get page content by ID
• confluence_get_page_by_title: Get page content by title within a space
• confluence_search: Search Confluence content using CQL

Security Considerations

• Credentials are stored locally in the env file and never leave your machine
• The server runs locally and communicates only with your Confluence instance
• API tokens can be revoked at any time from your Atlassian account