A lightweight MCP server that crawls Confluence subtrees and serves cached Markdown.
Why it exists
The crawler walks a graph and not just a tree: child pages, linked Confluence pages, and external links one level deep. A single hub page can pull in several subtrees.
What it does
search_pages- Keyword search across cached pages, with title weighted 10x.
get_page- Full Markdown content by page ID.
list_pages- An index of all cached pages.
refresh_cache- Force a re-crawl and bypass the TTL.
In-page directive
[confluence-mcp]
max_depth=3
follow_links=false
follow_external=true