# KS-GEN-AI/confluence-mcp-server [Health: Active]

**Category:** 🛠️ Other Tools and Integrations  
**Repository:** https://github.com/KS-GEN-AI/confluence-mcp-server  
**GitHub Stars:** 14  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/ks-gen-ai-confluence-mcp-server

## Description
Get Confluence data via CQL and read pages.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "confluence-mcp-server": {
    "command": "npx",
    "args": ["-y","@smithery/cli"],
    "env": {
      "CONFLUENCE_URL": "",
      "CONFLUENCE_API_MAIL": "",
      "CONFLUENCE_API_KEY": ""
    }
  }
}
```

**Requires environment variables:** `CONFLUENCE_URL`, `CONFLUENCE_API_MAIL`, `CONFLUENCE_API_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What KS-GEN-AI/confluence-mcp-server MCP server does

The KS-GEN-AI/confluence-mcp-server MCP server connects an MCP client to Confluence operations for searching, reading, and updating pages. It provides three tools:

- `execute_cql_search` runs a Confluence Query Language (CQL) query and accepts a result limit, which defaults to 10.
- `get_page_content` retrieves the content of a page using its `pageId`.
- `update_page_content` changes a page's content using its `pageId` and `content`. A `title` can also be supplied when the page title should change.

This makes the server suitable for assistants that need to locate relevant Confluence pages before reading them, or that need to write revised content back to an existing page.

## How it works

The KS-GEN-AI/confluence-mcp-server MCP server is implemented in TypeScript and communicates with MCP clients over standard input and output. The client starts the built Node.js entry point, and the server uses environment variables to connect to the configured Confluence instance.

A typical interaction can begin with `execute_cql_search`, using a CQL expression and an optional limit. The client can then pass a returned page identifier to `get_page_content`. For editing workflows, `update_page_content` accepts the target identifier and replacement content, with an optional new title.

Because the tools operate on page identifiers and CQL queries, callers need to provide valid Confluence-specific values. The README does not document additional filtering, pagination behavior, or operations for creating or deleting pages.

## Setup and configuration

To build the KS-GEN-AI/confluence-mcp-server MCP server locally, install the project dependencies with `npm install`, then run `npm run build`. The repository also provides `npm run watch` for development with automatic rebuilding.

Claude Desktop can start the compiled server with Node.js. Its MCP configuration supplies these environment variables:

- `CONFLUENCE_URL`: the Confluence site URL, such as an Atlassian wiki endpoint.
- `CONFLUENCE_API_MAIL`: the email address associated with the Confluence API credentials.
- `CONFLUENCE_API_KEY`: the Confluence API key.

The README points to Atlassian's profile security page as the location for obtaining an API token. The configured command uses the project's built `build/index.js` file, so the project must be built before Claude Desktop starts it.

## Tools and capabilities

The KS-GEN-AI/confluence-mcp-server MCP server covers a focused set of page workflows:

1. Search pages by submitting CQL through `execute_cql_search`.
2. Set a maximum number of search results with the `limit` parameter.
3. Fetch a page's content by ID.
4. Replace page content and optionally change its title.

The server is intended for existing Confluence content. The documented tools do not include a page-creation tool, attachment management, space administration, or broader Confluence configuration.

## Limitations and notes

The README documents Claude Desktop setup but does not list configuration examples for other MCP clients. It also does not state a license, package publishing workflow, or a hosted endpoint. The server requires Confluence credentials supplied through environment variables, and the API key must be provided by the user. For local troubleshooting, the project includes an `npm run inspector` script that launches the MCP Inspector workflow for its stdio connection.

_Full upstream README: https://allmcps.com/mcp/ks-gen-ai-confluence-mcp-server/readme_