Atlassian Confluence MCP Server
io.github.anuragagrawal0430/atlassian-confluence-mcp-server
Overview
STDIO MCP server for Atlassian Confluence Cloud, Server, and Data Center with 42 tools for spaces, pages, search, labels, comments, attachments, permissions, watchers, tasks, and exports.
Documentation
Atlassian Confluence MCP Server
Open-source MCP server for Atlassian Confluence. It lets MCP clients read, create, search, and manage Confluence wiki content across Confluence Cloud, on-premise Server, and Data Center deployments.
Installation
Run without installing:
npx atlassian-confluence-mcp-server
The documented MCP client launch uses:
{ "command": "npx", "args": ["-y", "atlassian-confluence-mcp-server"] }
You can also install globally with npm install -g atlassian-confluence-mcp-server and run atlassian-confluence-mcp-server, or build from source with npm install and npm run build. package.json declares Node.js >=20.0.0; the README prerequisite says Node.js 18+.
Configuration
Configuration is read from environment variables, with dotenv loading a local .env file. CONFLUENCE_BASE_URL is required. Authentication requires either PAT, or both CONFLUENCE_USERNAME and CONFLUENCE_PASSWORD.
For Confluence Cloud, set CONFLUENCE_BASE_URL, CONFLUENCE_USERNAME, and PAT. The client detects atlassian.net hosts and uses Basic auth with email plus API token. For Confluence Server/Data Center 7.9+, set CONFLUENCE_BASE_URL and PAT; Bearer auth is used. For older Server versions without PAT support, set CONFLUENCE_BASE_URL, CONFLUENCE_USERNAME, and CONFLUENCE_PASSWORD for Basic auth.
Secret and user-specific values are intentionally empty in this registry entry.
Capabilities
The current source registers 42 tools: confluence_add_page_comment, confluence_add_page_label, confluence_copy_page, confluence_create_page, confluence_create_page_in_personal_space, confluence_create_private_space, confluence_create_space, confluence_delete_page, confluence_delete_page_label, confluence_delete_space, confluence_export_page, confluence_get_child_pages, confluence_get_comment_attachments, confluence_get_current_user, confluence_get_page, confluence_get_page_ancestors, confluence_get_page_attachments, confluence_get_page_by_title, confluence_get_page_by_url, confluence_get_page_comments, confluence_get_page_labels, confluence_get_page_permissions, confluence_get_page_properties, confluence_get_page_tasks, confluence_get_page_version, confluence_get_page_versions, confluence_get_page_watchers, confluence_get_pages, confluence_get_personal_space_key, confluence_get_recently_modified, confluence_get_space, confluence_get_space_by_key, confluence_get_space_homepage, confluence_get_spaces, confluence_move_page, confluence_search, confluence_search_pages, confluence_set_page_permissions, confluence_test_connection, confluence_unwatch_page, confluence_update_page, confluence_watch_page.
The server also registers one resource, confluence://spaces, which returns the list of Confluence spaces as JSON. Features reviewed include space management, page CRUD, search, labels, comments, attachments, page hierarchy, version history, properties, current user lookup, watchers, permissions, recently modified pages, inline tasks, export URL generation, URL parsing, and personal/private space helpers.
Limitations
The server is STDIO-only in the current implementation. It does not expose HTTP or remote transport. It intentionally makes network requests to the configured Confluence instance. Request timeout is 30 seconds, pagination limits are clamped to 100, dynamic path segments are encoded, and error messages are sanitized and truncated. Credentials are environment-only and are not hard-coded.
Repository: https://github.com/anuragagrawal0430/atlassian-confluence-mcp-server