mcp -atlassian-confluence

This MCP server connects Claude, Cursor AI, and other AI assistants directly to your Atlassian Confluence knowledge base. It exposes a set of generic Confluence API tools that let you read data, create resources, and update content from your Confluence instance through a unified MCP interface. The server supports five tools (conf_get, conf_post, conf_put, conf_patch, conf_delete) that map to standard HTTP methods, enabling you to browse spaces, pages, and related content, or perform actions like creating pages or updating page content via a simple, consistent request format. You can run commands via npx (as shown in the Quick Start) or install the MCP server globally for integration with your preferred AI assistant. When using Claude or other helpers, configure the MCP server as a standard MCP endpoint, typically with a transport that your assistant supports (STDIO, HTTP, etc.).

Prerequisites:

• Node.js and npm installed on your system
• Access to a Confluence Cloud/Server instance with API tokens

Option A: Run via npx (no local install)

1. Ensure Node.js and npm are installed.
2. Run the MCP server directly with npx (replace placeholders with real values):

export ATLASSIAN_SITE_NAME=your-company
export ATLASSIAN_USER_EMAIL=your.email@company.com
export ATLASSIAN_API_TOKEN=your_api_token

npx -y @aashari/mcp-server-atlassian-confluence

This starts the server and exposes the MCP endpoints according to the package defaults.

Option B: Install globally for repeated use

1. Install the MCP server globally:

npm install -g @aashari/mcp-server-atlassian-confluence

2. Run the server (the binary is typically named something like mcp-atlassian-confluence):

export ATLASSIAN_SITE_NAME=your-company
export ATLASSIAN_USER_EMAIL=your.email@company.com
export ATLASSIAN_API_TOKEN=your_api_token

mcp-atlassian-confluence

Option C: Local project usage

1. In your project directory, install as a dev dependency:

npm install --save @aashari/mcp-server-atlassian-confluence

2. Run via npx or a script in your package.json that starts the MCP server with the required env vars.

Prerequisites recap:

• Node.js v14+ (prefer latest LTS)
• An Atlassian site name, user email, and API token configured as environment variables or in a .env file

Tips and common issues:

• Environment variables: Atlassian credentials can be provided via environment variables, a .env file, or a system-wide MCP config. Ensure ATLASSIAN_SITE_NAME matches yourcompany.atlassian.net, and that the API token has the necessary scopes for the endpoints you plan to use.
• Rate limits: Confluence Cloud API tokens can have rate limits. If you hit throttling, implement backoff in your AI tooling or stagger requests.
• Output format: The MCP tools support TOON by default for token efficiency. If you need standard JSON for downstream tooling, set outputFormat: "json" in your requests.
• Security: Never commit API tokens to source control. Use environment variables or secret stores in production.
• Endpoints: The tools expose generic endpoints (GET, POST, PUT, PATCH, DELETE). For examples, use the documented /wiki/api/v2/spaces, /wiki/api/v2/pages, and /wiki/rest/api/search paths.
• Config flexibility: You can place credentials under different keys in the config file (e.g., confluence, atlassian-confluence, or the package name). The system accepts multiple alias keys.