Knowtis

What the Knowtis MCP server does

The Knowtis MCP server gives AI assistants access to collaborative notes stored in Knowtis. Its documented purpose is to let assistants create, search, and manage notes. This makes it relevant for workflows where an agent needs to find existing material, add new notes, or maintain note content through an MCP-compatible client.

Knowtis itself is a local-first notes workspace with real-time collaboration. The application includes a document editor, AI-assisted editing, voice-note transcription, and study features such as flashcards, quizzes, summaries, and mind maps. The MCP server focuses on exposing the notes workspace to external AI assistants; the excerpt does not specify that every application feature is available as an MCP operation.

How it works

The repository contains a standalone MCP application under apps/mcp/, implemented with Hono. The wider application uses a NestJS API, PostgreSQL, Redis, and real-time collaboration components, but the excerpt does not describe the MCP transport, authentication flow, endpoint format, or individual tool schemas.

When connected to a compatible client, the Knowtis MCP server makes note operations available to the assistant. The README specifically names Claude Desktop, Cursor, and VS Code as clients that can connect to it. The current material does not state whether the server runs over standard input/output, HTTP, or another transport, so client configuration should follow the project’s dedicated MCP documentation.

Setup and configuration

Running the full Knowtis project requires Node.js 22, pnpm 10, and Docker. The documented setup command installs dependencies, creates environment files, starts PostgreSQL and Redis, and runs migrations:

pnpm run setup

Development services can then be started with:

pnpm dev:all

The README lists the API on port 3333, the notes application on 4200, and the backoffice application on 4400. These commands describe the full workspace rather than a confirmed MCP-only launch procedure.

AI features require OPENROUTER_API_KEY in apps/api/.env and the ai_enabled feature flag. That requirement applies to the application’s AI features; the excerpt does not identify an environment variable required specifically by the Knowtis MCP server. Instructions for connecting an MCP client are referenced in docs/LOCAL_SETUP.md and docs/MCP.md, but their contents are not included here.

Tools and capabilities

The provided material supports these MCP use cases:

• Create collaborative Knowtis notes.
• Search Knowtis notes.
• Manage existing Knowtis notes.
• Connect note workflows to Claude Desktop, Cursor, or VS Code.

The README does not provide formal tool names, argument definitions, return schemas, permission details, or examples of tool calls. Integrations with Anthropic, OpenAI, and Google are described for Knowtis’s AI system, not explicitly as MCP-client integrations.

Limitations and notes

The Knowtis MCP server should be evaluated against the exact note operations your agent requires. The supplied documentation does not establish support for flashcard generation, quizzes, mind maps, voice transcription, real-time cursors, or offline synchronization through MCP. Those capabilities belong to the broader Knowtis application and are not confirmed as exposed tools.

The repository is licensed under MIT. The available excerpt also does not specify deployment instructions for the standalone MCP application, authentication and authorization behavior, network requirements, or whether a running Knowtis instance is required. Consult the repository’s MCP and local-setup documentation before configuring the Knowtis MCP server in a production or shared environment.