This MCP server provides two tools for interacting with a Confluence instance: search_confluence and fetch_confluence_page. The search tool translates natural language queries into Confluence CQL searches, supports optional filters like spaces, labels, and date ranges, and returns a ranked list with titles, URLs, spaces, excerpts, and labels. The fetch tool hydrates a specific page by its content_id, returning a Markdown document with metadata (URL, space, version, last-modified date, labels, and ancestors) followed by the full page body. You can connect via stdio-based clients (e.g., Claude Desktop, Cursor) or HTTP-based transports (e.g., Windsurf) by running the server in the appropriate mode.

To use the stdio workflow with Claude/Desktop clients, start the MCP server using uv and then configure the client to send commands over stdio. For HTTP-based clients, run the server in a server mode (e.g., via Docker) and point the client to http://<host>:<port>/mcp. The provided configuration shows how to wire the uv-based stdio runtime for the confluence server and supply necessary environment variables such as CONFLUENCE_URL and CONFLUENCE_PERSONAL_ACCESS_TOKEN.

Prerequisites

• Python 3.11+ and UV runtime (uv) installed on your system, or Docker for container deployment
• A Confluence instance accessible over HTTP(S)
• A Confluence Personal Access Token with appropriate permissions

Step 1: Prepare environment

• Install Python 3.11+ (from python.org or your OS package manager)
• Install the UV runtime following the project’s guidance (per the repository’s prerequisites)

Step 2: Obtain and configure credentials

• Create a .env file (or equivalent) with: CONFLUENCE_URL=https://confluence.example.com CONFLUENCE_PERSONAL_ACCESS_TOKEN=your_token_here Optional TLS/verification envs may be added as needed, e.g., CONFLUENCE_VERIFY_TLS=true

Step 3: Install dependencies (if applicable)

• If the project uses a virtual environment: python -m venv venv source venv/bin/activate # On Windows use venv\Scripts\activate
• Install project dependencies (adjust to the project’s setup, e.g., poetry, pip requirements): pip install -r requirements.txt # if a requirements file exists

Step 4: Run the MCP server

• Quickstart (stdio mode for Claude Desktop / Cursor): uv sync uv run python -m confluence_search.fastmcp_app

• Docker (HTTP transport): python deploy_confluence_mcp.py # builds image + starts container on :43043

  or manually:

  docker build -t confluence-mcp-server . docker run -d --name confluence-mcp-server
  --env-file .env
  -p 43043:43043
  confluence-mcp-server

• If using Docker, connect clients to http://127.0.0.1:43043/mcp

Environment variables drive runtime behavior. Ensure CONFLUENCE_URL and CONFLUENCE_PERSONAL_ACCESS_TOKEN are kept secret. The server supports a read-only mode for safety. The cache directory (CONFLUENCE_CACHE_DIR) defaults to .cache/confluence_mcp and can be customized. When using the HTTP transport, the port defaults to 43043 unless changed by environment variables. If you encounter TLS issues, verify your TLS settings (CONFLUENCE_VERIFY_TLS) and token permissions. The npm_package field is null for this Python-based MCP server.