Add the KnowMan MCP server to LM Studio

What you need

LM Studio 0.3.17 and later acts as an MCP Host, so it can connect MCP servers to the app and make them available to your models (LM Studio docs). To run the local KnowMan server you need Node 22 or newer (KnowMan MCP).

You also need a KnowMan API key. Create one in the web app at Settings > Connections > Connect an AI agent. Keys look like km_... and are shown exactly once, so copy the key right away (KnowMan MCP).

Add the hosted KnowMan server

Switch to the "Program" tab in the right-hand sidebar and click Install > Edit mcp.json. This opens mcp.json in the in-app editor, where you add MCP servers by editing the file (LM Studio docs).

Paste the snippet from the config block above and replace km_... with your real key. LM Studio's mcp.json follows Cursor's notation, and remote servers are configured with a url plus a headers object containing the Authorization bearer value (LM Studio docs). The hosted KnowMan server lives at https://api.knowman.io/mcp (KnowMan MCP).

If you add a server manually next to an existing one, copy only the content after "mcpServers": { and before the closing brace, so the file stays valid (LM Studio docs).

Run KnowMan as a local server instead

LM Studio 0.3.17 (b10) supports both local and remote MCP servers (LM Studio docs). If you prefer the local stdio server, use the same mcp.json editor with this entry, which follows KnowMan's documented stdio config (KnowMan MCP):

```json { "mcpServers": { "knowman": { "command": "npx", "args": ["-y", "@knowman/mcp"], "env": { "KNOWMAN_API_KEY": "km_..." } } } } ```

This needs Node 22 or newer on the machine running LM Studio (KnowMan MCP).

Sign in once with knowman login

For the local setup you can skip pasting the key into mcp.json. Run npx -p @knowman/mcp knowman login once and sign in in the browser. The saved credential is shared by the CLI and the MCP server, so the env block with KNOWMAN_API_KEY can be dropped from the config (KnowMan MCP).

The same package installs a knowman CLI with commands such as kb-search, ask, save, recent, memories, status and logout, which is handy for checking the account the server uses (KnowMan MCP).

Check that KnowMan answers

Once the server entry is in mcp.json, connected MCP servers become available to your models (LM Studio docs). The simplest check is to ask your model something concrete, for example "search my KnowMan for reading lists" or "save this URL" (KnowMan MCP).

If nothing happens, retrace the two inputs: the server entry in mcp.json and, for the hosted server, a valid key in the Authorization header. If the model reports a rejected key, recreate the key at Settings > Connections in the web app (KnowMan MCP). For more background on the protocol, see what an MCP server is.

The six KnowMan tools

KnowMan exposes six tools to MCP clients (KnowMan MCP):

kb_search searches saved items by keyword or meaning and returns titles, URLs and metadata. kb_ask answers a question with citations. kb_get_item fetches the readable content of one item. kb_save saves a URL to the Inbox or Knowledge Base with an optional note. kb_recent lists recent items. kb_memories lists saved claims, decisions and playbooks, and is read-only (KnowMan MCP).

Token use and trust with local models

LM Studio warns that some MCP servers designed for Claude, ChatGPT or Gemini may use excessive amounts of tokens, which can bog down a local model and trigger frequent context overflows (LM Studio docs). This applies to any server you add, not only KnowMan.

The same doc advises never installing MCP servers from untrusted sources, since some can run arbitrary code, access local files and use your network connection (LM Studio docs). The KnowMan server is open source under MIT, so the code can be inspected before you run it (KnowMan MCP). For a wider picture, see how to connect any AI to your notes.