Confluence

You can run a Confluence MCP Server to bridge Confluence content with Large Language Models. It provides search, read, and listing capabilities so your LLMs can query Confluence pages, fetch page content in Markdown, and explore space content efficiently, all through the MCP protocol.

How to use

Use an MCP client to connect to the Confluence MCP Server in either stdio (local) mode or HTTP/SSE (containerized/remote) mode. You can search for content across your Confluence instance, fetch full page content converted to Markdown, and list all pages within a space. The server handles authentication errors gracefully and converts HTML content to clean Markdown for optimal context usage.

How to install

Prerequisites you need before installing the MCP server are Python 3.11 or higher, uvx (recommended) or pip, Confluence Cloud or Server, and a Confluence API token. Optionally, Docker is available for containerized deployment.

Additional sections

Configuration and deployment details are provided below. You will learn how to run the server locally in stdio mode, run it in HTTP/SSE mode via a Docker container, and how to connect using an MCP Inspector for testing and debugging.

Available tools

search_confluence

Search across Confluence content for pages, blog posts, and attachments and return matching results with IDs and types.

read_page

Fetch the full content of a Confluence page and convert HTML to Markdown for better LLM context usage.

list_space_content

List all pages within a specified Confluence space, providing page titles and IDs and total counts.