MCP Server Atlassian Confluence

What aashari/mcp-server-atlassian-confluence MCP server does

The aashari/mcp-server-atlassian-confluence MCP server exposes Confluence Cloud through a small set of generic tools rather than separate tools for every resource type. It can work with spaces, pages, page bodies, blog posts, labels, comments, attachments, and other endpoints available through the Confluence API.

The read operation supports common Confluence paths such as /wiki/api/v2/spaces, /wiki/api/v2/pages, page body endpoints, and /wiki/rest/api/search with CQL. Write operations can create pages and blog posts, add labels or comments, and update or remove resources. Page content can be converted from Atlassian Document Format to Markdown.

This fits the Workplace & Productivity category for assistants that need access to team documentation, project knowledge, policies, specifications, or support material.

How it works

The server provides five MCP tools:

• conf_get sends GET requests to Confluence.
• conf_post creates resources with POST requests.
• conf_put replaces resources with PUT requests.
• conf_patch applies partial updates with PATCH requests.
• conf_delete removes resources with DELETE requests.

Each tool accepts an endpoint path. Optional queryParams pass parameters such as limit, space-id, body-format, or cql. Write tools also require a JSON body.

Responses are TOON by default, which the project describes as using 30–60% fewer tokens than JSON. Set outputFormat to json when standard JSON is more useful. Use the jq parameter to select only required fields. A practical discovery pattern is to request one result without filtering, inspect its fields, and then add a focused expression such as results[*].{id: id, title: title}. Limiting result counts and filtering fields helps avoid unnecessarily large responses.

Setup and configuration

The aashari/mcp-server-atlassian-confluence MCP server uses an Atlassian site name, user email, and API token. The site name is the subdomain in an address such as your-company.atlassian.net.

Run it with npx -y @aashari/mcp-server-atlassian-confluence, or install the package globally with npm. The resulting mcp-atlassian-confluence binary can be configured for an assistant using STDIO transport. Claude Desktop configuration is documented using the command, args, and env fields.

Credentials may come from environment variables, a .env file in the current directory, or ~/.mcp/configs.json. The supported credential variables are ATLASSIAN_SITE_NAME, ATLASSIAN_USER_EMAIL, and ATLASSIAN_API_TOKEN.

Tools and capabilities

The aashari/mcp-server-atlassian-confluence MCP server can list and inspect spaces, retrieve pages and page bodies, find child pages and labels, search content with CQL, and access blog posts. It can also create pages or blog posts, add labels and comments, replace pages or blog posts, partially update spaces or comments, and delete pages, blog posts, labels, comments, or attachments where the relevant endpoint is available.

Confluence v2 updates generally require the complete resource for PUT requests, and page updates require an incremented version number. Most DELETE endpoints return no content on success.

Limitations and notes

The tools map closely to Confluence API paths, so callers need the correct endpoint, parameters, request body, and resource identifiers. Unfiltered responses may contain substantial metadata and increase token use. PUT replaces the full resource rather than changing only selected fields.