Knock MCP Server
Manage notification workflows and infrastructure directly from your AI agent interface.
Data last scanned today · Reviewed today
Overview
Knock MCP Server acts as a bridge between AI assistants and the Knock notification platform. It provides tools to interact with workflows, templates, user data, and environment logs, utilizing the Knock Management API to allow developers to inspect and modify their notification infrastructure without leaving their coding environment.
Our verdict
With a security score of 35/100, this server is currently classified as high-risk. This rating reflects the fact that the project is maintained by the community rather than the official vendor and has not yet undergone an editorial security review. Given the lack of independent oversight and the potential for write access, it is recommended primarily for users comfortable managing their own integration security in non-production environments.
- TypeScript codebase with twelve direct dependencies
- Provides documented access to the Knock Management API
- Includes OAuth 2.1 + PKCE authentication support
- High-risk security score of 35/100
- Community-maintained rather than official vendor support
- No current editorial review of read-only mode safety
Setup
{
"mcpServers": {
"knock-mcp": {
"args": [
"-y",
"@knocklabs/mcp"
],
"command": "npx"
}
}
}Paste into ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows), then restart Claude Desktop. Some servers require API keys or tokens as environment variables — check the project's README and add an "env" object if needed.
- No read-only mode — running this grants full read/write access.
- No built-in authentication — anyone with access to this config can invoke every tool it exposes.
Tools
| Tool | Description | Risk |
|---|---|---|
| search_mapi | Search the Knock Management API. | low |
| execute_mapi_read | Execute GET requests against the Knock Management API. | low |
| execute_mapi_write | Execute POST, PUT, PATCH, and DELETE requests against the Knock Management API. | high |
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| ChatGPT | ||||
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf |
Frequently asked questions
›How do I authenticate for interactive use?
Interactive clients use OAuth 2.1 + PKCE via Knock's AuthKit. Your browser will open on the first connection to authorize the session and allow you to select specific tool capabilities.
›Can I use this server in CI or headless environments?
Yes, you can authenticate via a Knock service token by adding an Authorization header containing your bearer token to your MCP configuration. This mode skips OAuth and enables all tool groups by default.
›Which tool groups are enabled by default?
By default, the server enables the Management API (code mode), Knock agent, Debug tools, and Documentation. Capabilities like managing data or classic resource management remain opt-in.
›What are the prerequisites for self-hosting the server?
Self-hosting requires Node.js 18+, a Cloudflare account with Workers enabled, and a Knock account configured with AuthKit.
›How does privilege management differ between OAuth and service tokens?
OAuth consent lets you choose specific capabilities during authorization, whereas service token sessions grant all tool groups, with privilege levels determined by the scopes attached to your specific Management API token.
Badge
Maintain this server? Add the live badge to your README.