Server data from the Official MCP Registry
KnowMan MCP server: search, ask, save and read your Knowledge Base from any MCP client.
About
KnowMan MCP server: search, ask, save and read your Knowledge Base from any MCP client.
Remote endpoints: streamable-http: https://api.knowman.io/mcp
Security Report
The server requires a KnowMan API key (env var, CLI flag, or a stored login file) and its network traffic goes to the documented api.knowman.io endpoint, which matches its purpose as a knowledge base client. The main items to be aware of are that search queries, questions, and saved URLs are sent to the remote KnowMan service by design, and the stored-credential file handling could not be fully verified because the cli-account.ts source is truncated. Supply chain analysis found 6 known vulnerabilities in dependencies (0 critical, 4 high severity). Package verification found 1 issue.
6 files analyzed · 12 issues found
Security scores are indicators to help you make informed decisions, not guarantees. Always review permissions before connecting any MCP server.
Permissions Required
This plugin requests these system permissions. Most are normal for its category.
What You'll Need
Set these up before or after installing:
Environment variable: KNOWMAN_API_KEY
How to Install & Connect
Available as Local & Remote
This plugin can run on your machine or connect to a hosted endpoint. during install.
Documentation
View on GitHubFrom the project's GitHub README.
@knowman/mcp
A stdio MCP server that lets an MCP client search, ask, save and read items in a KnowMan account.
Tools
Hosted server (remote)
| Tool | Purpose |
|---|---|
kb_research | Retrieve ranked passages and memories across the whole library in one call; the assistant writes the answer. |
kb_get_item | Read content and metadata for one item. |
kb_recent | List recently saved items. |
kb_save | Save a URL to Inbox or Knowledge Base. |
Local stdio package
| Tool | Purpose |
|---|---|
kb_research | Retrieve ranked passages and memories across the whole library in one call; the assistant writes the answer. Also served by the local package from 0.3.1. |
kb_get_item | Read content and metadata for one item. |
kb_save | Save a URL to Inbox or Knowledge Base. |
kb_recent | List recently saved items. |
kb_search | Search saved items without bodies; results carry trust. Deprecated; use kb_research on the hosted server. |
kb_ask | Ask a question over the whole library (Inbox and Knowledge Base) by default; an optional scope narrows it to either location. Deprecated; use kb_research on the hosted server. |
kb_memories | List saved memories such as claims, decisions and playbooks; read-only. Deprecated; use kb_research on the hosted server. |
Install
npx -y @knowman/mcp runs the server on demand. npm install -g @knowman/mcp installs the knowman-mcp command.
Requires Node.js 22 or newer.
Get a key
Sign in at https://app.knowman.io, create an API key in settings, and set it as KNOWMAN_API_KEY. Never commit the key.
Configuration
Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"knowman": {
"command": "npx",
"args": ["-y", "@knowman/mcp"],
"env": {
"KNOWMAN_API_KEY": "<your-key>"
}
}
}
}
Claude Code:
claude mcp add knowman --env KNOWMAN_API_KEY=<your-key> -- npx -y @knowman/mcp
Cursor (~/.cursor/mcp.json, or .cursor/mcp.json in a project): use the same mcpServers JSON as the Claude Desktop example above.
Prefer signing in with the browser instead of pasting a key? Install the package globally (npm install -g @knowman/mcp) and run knowman login. The saved credential is shared with knowman-mcp, so the env block is not needed. Run knowman status to check it.
The package runs on your own machine over stdio. KnowMan also runs a hosted server at the endpoint in the next section, for clients that only accept remote MCP servers, such as claude.ai custom connectors.
Remote (hosted) server
Endpoint: https://api.knowman.io/mcp (Streamable HTTP, stateless, JSON responses, MCP protocol 2025-11-25). It exposes four tools: kb_research, kb_get_item, kb_recent and kb_save (see Tools above).
claude.ai. Open Settings, then Connectors, then Add custom connector, and enter https://api.knowman.io/mcp. Sign in with your KnowMan account when the browser opens.
Sign in with OAuth. The server implements OAuth 2.1 for MCP: protected resource metadata at /.well-known/oauth-protected-resource/mcp, authorization server metadata at /.well-known/oauth-authorization-server, dynamic client registration, authorization code with PKCE (S256), and rotating refresh tokens. A client that supports MCP OAuth connects from the endpoint URL alone: it opens a consent page at app.knowman.io, you sign in and approve, and the client receives a one-hour access token plus a refresh token. Access is limited to two scopes: knowman.read (search and read) and knowman.write (save links).
Or use an API key. Create a key in your KnowMan settings and send it as Authorization: Bearer <key>:
{
"mcpServers": {
"knowman": {
"url": "https://api.knowman.io/mcp",
"headers": { "Authorization": "Bearer <your-key>" }
}
}
}
Claude Code: claude mcp add --transport http knowman https://api.knowman.io/mcp --header "Authorization: Bearer <your-key>".
What has been verified. The official MCP TypeScript SDK client completes the whole OAuth flow against the live server (registration, consent, token, tool calls, refresh). We have not yet tested ChatGPT connectors or Cursor ourselves; the endpoint follows the requirements those products document, and the redirect URIs they use are allowed. If one of them fails to connect, please open an issue with the product and the error you see.
Disconnect an app. Open the API keys list in your KnowMan settings and delete the entry named OAuth: <app name>. Deleting it revokes the connection and its refresh token. Changing your password also revokes every connected app.
Limits. An API key or an OAuth token cannot create other API keys. Requests are rate limited per token and per IP.
CLI reference
KnowMan CLI
Usage:
knowman <command> [options]
Account:
login [--web-url <url>] [--api-url <url>] [--label <name>] [--no-browser] [--scopes agent|read]
Sign in through your browser; the credential is shared with knowman-mcp.
logout
Revoke the saved key on the server and remove it.
status
Show whether you are signed in (never prints the key).
Global flags:
--json
--format <json|text> json = one compact document; lists become {"results":[...]}
--api-url <url>
--api-key <key>
--help
--version
Commands (each also works without the kb- prefix, e.g. "save", "search"):
kb-research
Retrieve ranked evidence across the user's whole library and distilled memories. Default tool for factual, strategic and how-to questions. Returns source passages, not an answer.
--question <STRING>
--max_passages <NUMBER> [optional]
kb-search
Deprecated: use kb_research. Search saved items in the KnowMan Inbox, Knowledge Base, or Archive without returning content bodies; a location filter costs one extra lookup per hit. Every result carries `trust`: 'internal' is the user's OWN knowledge (their pages, documents and designated links) and outranks 'external' saved articles on anything about the user themselves. Returns snippets only; kb_get_item reads the full text.
--query <STRING>
Text to search for in KnowMan.
--limit <NUMBER> [optional]
Maximum number of search results to return.
--location <STRING> [optional] (one of: inbox|kb|archive)
Optional Inbox, Knowledge Base, or Archive location filter.
--scope <STRING> [optional] (one of: internal|external|all)
Which knowledge to search: 'internal' (the user's own pages, documents and designated links), 'external' (articles they saved), or 'all'. Defaults to all.
kb-ask
Deprecated: use kb_research. Ask a question over the whole KnowMan library by default (Inbox and Knowledge Base; pass scope to narrow to kb or inbox), or a single item when itemId is given; returns an answer plus citations whose itemId resolves via kb_get_item. Answers specific questions with citations; may abstain on broad questions: then use kb_memories, kb_search and kb_get_item.
--question <STRING>
Question to ask over the KnowMan Knowledge Base.
--item-id <STRING> [optional]
Optional item id restricting the question.
--project-id <STRING> [optional]
Optional project id restricting the question.
--scope <STRING> [optional] (one of: all|kb|inbox)
Which part of the library to search: all (default): the whole library, Inbox and Knowledge Base; kb: Knowledge Base only; inbox: Inbox only. Ignored when itemId is given.
kb-get-item
Read the full text of one saved item by id when research evidence needs more context.
--id <STRING>
KnowMan item id to retrieve.
kb-save
Save a URL to the KnowMan Inbox or directly to the Knowledge Base. Pass trust:'internal' when the link is the user's own knowledge (their company docs, their own writing) rather than an outside article.
--url <STRING>
URL to save into KnowMan.
--location <STRING> [optional] (one of: inbox|kb)
Save into the Inbox or directly into the Knowledge Base.
--note <STRING> [optional]
Optional note to attach to the saved item.
--trust <STRING> [optional] (one of: internal|external)
Mark the saved link as the user's own knowledge ('internal' — their company docs, their own posts) rather than an outside article ('external', the default).
kb-recent
List recently saved items. Use only when the user asks about recent saves.
--limit <NUMBER> [optional]
Maximum number of recent items to return.
--location <STRING> [optional] (one of: inbox|kb)
Optional Inbox or Knowledge Base location filter.
--scope <STRING> [optional] (one of: internal|external|all)
Which knowledge to list: 'internal', 'external', or 'all' (default).
--source-type <ARRAY> [optional]
Optional source type filter: manual, extension, rss, newsletter, inbox, or import. Array or comma-separated string. Use sourceType:['manual','extension'] to return only the user's own saves (manual and extension saves). Defaults to all types.
kb-memories
Deprecated: use kb_research. List saved memories (claims, decisions, playbooks) from the Knowledge Base; use kind 'playbook' to find step-by-step processes the user has taught KnowMan. Read-only. Best first stop for process, strategy and how-to questions (kind 'playbook').
--project <STRING> [optional]
Optional project id as a UUID; project name lookup is unavailable.
--kind <STRING> [optional] (one of: claim|decision|playbook)
Optional memory kind: claim, decision, or playbook.
--query <STRING> [optional]
Optional case-insensitive title and body substring filter.
--limit <NUMBER> [optional]
Maximum number of memories to return.
Reviews
No reviews yet
Be the first to review this server!
More Developer Tools MCP Servers
Rea
Freeby Morluto · Developer Tools
Reverse engineer anything from your terminal or agent with one CLI and MCP server.
Paperclip
Freeby Paperclipai · Developer Tools
Trending hip-hop artist momentum scores across four cultural dimensions.
Scrapling
Freeby D4Vinci · Developer Tools
Web scraping with stealth HTTP, real browsers, and Cloudflare bypass. CSS selectors supported.
Git
Freeby Modelcontextprotocol · Developer Tools
Read, search, and manipulate Git repositories programmatically
Toleno
Freeby Toleno · Developer Tools
Toleno Network MCP Server — Manage your Toleno mining account with Claude AI using natural language.
MCP Marketplace
Freeby mcp-marketplace · Developer Tools
Search and install MCP servers from inside your AI client.
