confluence-dc-mcp 0.4.0

MCP server for Confluence Data Center raw storage operations

Confluence Data Center MCP

Typed Python MCP server for Confluence Data Center raw storage operations.

Requirements

• Python 3.11+
• uv
• Confluence Data Center REST API access
• Either a Confluence Personal Access Token or username/password auth

Setup

uv sync
cp .env.example .env

Set the values in .env, then run:

set -a
source .env
set +a
uv run confluence-dc-mcp

Run with uvx

After the package is published, it can be run without cloning the repository:

CONFLUENCE_BASE_URL="https://confluence.example.com" \
CONFLUENCE_PAT="replace-with-token" \
uvx confluence-dc-mcp

Or with username/password authentication:

CONFLUENCE_BASE_URL="https://confluence.example.com" \
CONFLUENCE_USERNAME="alice" \
CONFLUENCE_PASSWORD="replace-with-password" \
uvx confluence-dc-mcp

For an MCP client using uvx with a Personal Access Token:

{
  "mcpServers": {
    "confluence-data-center": {
      "command": "uvx",
      "args": ["confluence-dc-mcp"],
      "env": {
        "CONFLUENCE_BASE_URL": "https://confluence.example.com",
        "CONFLUENCE_PAT": "replace-with-token"
      }
    }
  }
}

For an MCP client using uvx with username/password authentication:

{
  "mcpServers": {
    "confluence-data-center": {
      "command": "uvx",
      "args": ["confluence-dc-mcp"],
      "env": {
        "CONFLUENCE_BASE_URL": "https://confluence.example.com",
        "CONFLUENCE_USERNAME": "alice",
        "CONFLUENCE_PASSWORD": "replace-with-password"
      }
    }
  }
}

For MCP clients that pass environment variables directly, use:

{
  "mcpServers": {
    "confluence-data-center": {
      "command": "uv",
      "args": [
        "--directory",
        "/Users/mikhail/Projects/mcp-atlassian-confluence-raw-storage",
        "run",
        "confluence-dc-mcp"
      ],
      "env": {
        "CONFLUENCE_BASE_URL": "https://confluence.example.com",
        "CONFLUENCE_PAT": "replace-with-token"
      }
    }
  }
}

Environment

CONFLUENCE_BASE_URL is required and must be the root Confluence URL.

Use one authentication mode:

• CONFLUENCE_PAT
• CONFLUENCE_USERNAME and CONFLUENCE_PASSWORD

Optional settings:

• CONFLUENCE_VERIFY_SSL, defaults to true
• CONFLUENCE_TIMEOUT_SECONDS, defaults to 30

Tools

• health_check: validates server configuration.
• search_pages: searches content with CQL and returns page summaries.
• get_page_storage: returns raw storage-format XHTML for a content ID.
• get_page_storage_by_title: returns raw storage-format XHTML for a unique page title in a space.
• get_page_children: returns direct child page summaries for a content ID.
• get_attachment_list: returns attachment metadata for a page content ID.
• download_attachment: downloads an attachment as an embedded MCP resource for LLM analysis.
• update_page_storage: replaces raw storage-format XHTML and increments the Confluence version.

Development

uv run ruff check .
uv run mypy
uv run pytest
uv build

CI and Release

GitHub Actions run linting, strict type checking, tests, and package build on push and pull requests.

Publishing runs when a GitHub Release is published. The workflow uses PyPI Trusted Publishing, so configure a PyPI trusted publisher for this repository with:

• workflow: publish.yml
• environment: pypi
• package name: confluence-dc-mcp