Labsco
MCP SERVER

Atlassian Confluence

by aashari

Call any Confluence REST path from the session — five verbs, with a jq filter on the response so a page listing does not arrive as a wall of JSON.

Team Wikis & Documentation PlatformsVerified
Summary
You stop waiting for someone to wrap the endpoint you happen to need.

Five verbs reach anything the Confluence API exposes, and the jq parameter decides how much of the answer is worth spending context on — a page list can come back as ids and titles instead of full bodies. The cost is discoverability: the tool names tell you nothing about what is possible, so the path has to come from Atlassian's documentation rather than from the server.

What it is

A pass-through client for the Confluence REST API: five tools mapping to GET, POST, PUT, PATCH and DELETE, each taking a path, query parameters and a body, and each returning TOON by default.

What you get
  • Reads against any Confluence path, including the v2 pages and blogposts endpoints
  • Resource creation, full replacement, and partial update as three distinct verbs
  • Deletion of pages, blogposts and other resources by path
  • A jq parameter on every call that narrows the response to the fields you name, before it reaches the model
  • TOON responses by default, with JSON available per call through outputFormat
Requirements

An Atlassian site name, the account email and an API token. npx on your PATH. Every call takes the REST path itself, so Atlassian's API reference is part of the setup.

Setup effort

One command plus a key — npx -y @aashari/mcp-server-atlassian-confluence, then supply credentials