Linkedmash — Claude Code MCP server: LinkedIn bookmarks
Linkedmash — a remote MCP server that lets your assistant work with your LinkedIn bookmarks: filters by author, labels and dates, keyword or meaning-based search.
no license 204 installs updated 2025-07-29
claude mcp add --transport http linkedmash https://linkedmash.run.toolsWhat Linkedmash does
Seven tools sit behind a single endpoint. list_bookmarks returns saved posts with pagination and filters for read state, media type, author, labels or a date window, and author names match loosely. search_bookmarks takes either a keyword query or a semantic one; get_bookmark_count tallies whatever the current filter matches. Labels are handled by list_labels, add_labels and remove_labels.
Connection is one HTTP URL, https://linkedmash.run.tools. From Claude the command reads claude mcp add --transport http linkedmash https://linkedmash.run.tools. Cursor and GitHub Copilot want a JSON block under mcpServers or servers instead.
Semantic search depends on setup_vector_store having been called. Run search_bookmarks before that and it fails with “Vector Store Not Setup”, leaving you with keyword matching. Both label tools also require a label_id or a label_name argument; an empty call goes nowhere.
What it helps with
- find every saved post from one author
- filter unread bookmarks by date range
- search saved posts by topic instead of exact wording
- attach labels to a batch of bookmarks
- count bookmarks matching a filter
Example prompts for the agent
- Show my unread LinkedIn bookmarks, newest first.
- Search my bookmarks for posts about MCP servers by meaning.
- Add the label
laterto these posts and tell me how many bookmarks carry it.
How to install Linkedmash
Platform CLI: remote server
claude mcp add --transport http linkedmash https://linkedmash.run.toolsPlatform CLI: remote server
codex mcp add linkedmash --url https://linkedmash.run.toolsConfig file
{
"mcpServers": {
"linkedmash": {
"url": "https://linkedmash.run.tools"
}
}
}Config file
{
"servers": {
"linkedmash": {
"type": "http",
"url": "https://linkedmash.run.tools"
}
}
}Platform CLI: remote server
gemini mcp add -t http linkedmash https://linkedmash.run.toolsStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Linkedmash
How do I install Linkedmash in Claude Code?
Run claude mcp add --transport http linkedmash https://linkedmash.run.tools. The server is remote, so there is no local process to keep alive.
What does Linkedmash do?
It gives your assistant access to your LinkedIn bookmarks: filtered lists, keyword and semantic search, counts, and label editing.
Why does semantic search fail with a vector store error?
The vector store has not been initialized yet. Call setup_vector_store first, then search_bookmarks can match on meaning.