getSpaceLABSCO SUMMARY
Two managers instead of twelve tools, which keeps the tool list short.
managePages and manageComments each take an action and validate only the parameters that action needs, so create, update, delete and read share one definition. The Markdown conversion matters more than it sounds: Confluence storage format is not Markdown, and writing it by hand is where most integrations produce broken pages.
What it is
A TypeScript server over the Confluence REST API. Eleven tools: two combined managers that take an action parameter, space, page, comment and search reads, and three Markdown export tools.
What you get
- Eight core tools — the README's count, consolidated down from twelve — plus three export tools, exportPage, exportPageHierarchy and batchExportPages, for 11 in all
- Pages created, updated, deleted, and fetched as metadata or full content, through one action parameter — managePages
- A page fetched precisely by its title — getPageByPrettyUrl, and a space's information — getSpace
- Comments created, updated, deleted and replied to, both ordinary and inline, through one manager — manageComments
- A page's comments with paging, and one comment in detail — getPageComments, getComment
- Content searched with CQL, falling back to a basic search when the CQL is malformed — searchContent, and comments searched within a space — searchComments
- Markdown detected and converted to HTML automatically on page and comment writes, plus Markdown export of page content
- HTTP keep-alive, response compression, a configurable request timeout and automatic retry on failure
Requirements
CONFLUENCE_URL, plus either CONFLUENCE_ACCESS_TOKEN, which the project recommends, or CONFLUENCE_USERNAME with CONFLUENCE_PASSWORD. Node with npm; you build it from the repository and it runs from dist/index.js. PORT, NODE_ENV, TIMEOUT and REJECT_UNAUTHORIZED are optional.
Setup effort
Build from source
USE CASES
01
Publishing something you wrote in Markdown
Create the page under a parent and let the server convert the Markdown to Confluence's own format rather than pasting raw text.
02
Finding a page you only know by title
Fetch it precisely by title, or search the space with CQL and fall back to a basic search when the query is malformed.
03
Working through review comments
List a page's comments with paging, read one in detail, and reply — inline comments included — through the same manager.
04
Updating a page without rewriting it
Fetch the current content, apply an incremental update, and let the conversion handle the formatting on the way back in.
11 tools
Listed in project docs. Your client may show a different list.
getPageByPrettyUrlmanagePagesmanageCommentsgetPageCommentsgetCommentsearchContentsearchCommentsexportPageexportPageHierarchybatchExportPagesBEFORE YOU INSTALL
What to know first
Tools you get
11 tools named in the project's documentation.
Requires
Node.js >= 14.0.0
INSTALLATION
Connect your client
MCP Clients
Cursor — local build
CURSOR — LOCAL BUILD · TERMINAL
git clone https://github.com/enjoyzl/mcp-server-confluence-ts
cd mcp-server-confluence-ts
npm install
npm run build:cleanAdd this block to ~/.cursor/mcp.json:
CURSOR — LOCAL BUILD · JSON
{
"mcpServers": {
"mcp-server-confluence-ts": {
"command": "node",
"args": [
"<YOUR_PATH_TO_MCP_SERVER_CONFLUENCE_TS>/dist/index.js"
],
"env": {
"CONFLUENCE_URL": "<YOUR_CONFLUENCE_URL>",
"CONFLUENCE_USERNAME": "<YOUR_CONFLUENCE_USERNAME>",
"CONFLUENCE_PASSWORD": "yourpwd"
}
}
}
}<YOUR_PATH_TO_MCP_SERVER_CONFLUENCE_TS> is the full path to your clone. To use an access token instead (the project recommends it), set CONFLUENCE_ACCESS_TOKEN instead of CONFLUENCE_USERNAME and CONFLUENCE_PASSWORD.
