Msty ​

Msty is a desktop AI studio that goes beyond simple chat. It features RAG-powered "knowledge stacks" that let you ground conversations in your own documents, side-by-side model comparison for evaluating different LLMs on the same prompt, and offline mode using local models. MCP support brings live external data into this mix.

Ms
Msty
msty.app · msty.app
TRANSPORT
Streamable HTTP ✓
PLATFORM
Windows · macOS · Linux
MCP VIA
Settings UI

Knowledge Stacks + Live MCP Data ​

Knowledge stacks are Msty's version of RAG. You create a "stack" from local documents — PDFs, code, Markdown — and the AI references them during conversations. MCP adds a real-time dimension: instead of only reading static documents, the model can query live APIs, databases, or monitoring tools.

For example, pair a stack of your team's runbooks with an MCP server that reads production metrics. Ask "what's the recovery procedure for high memory usage, and is it happening now?" — Msty answers from both the runbook and the live data.

Unique features:

  • Knowledge stacks — document-grounded RAG with vector search
  • Model comparison — run the same prompt across Claude, GPT, Gemini, and Ollama side by side
  • Offline mode — Ollama integration for fully local operation
  • Chat organization — folders, tags, and search for past conversations
  • Export — conversations export as Markdown or JSON
  • Cross-platform — native desktop app for all operating systems

Adding Vinkius Cloud ​

1. Get a Token ​

In Vinkius Cloud, open your server → Connection Tokens → Create. Copy the URL.

2. Configure in Msty ​

Open Msty → Settings → MCP Servers → Add Server. Paste your URL:

https://edge.vinkius.com/{TOKEN}/mcp

3. Use in Chat ​

Start or continue a conversation. MCP tools are available alongside your knowledge stacks. Msty decides automatically when to use each.

FAQ ​

Can Msty use knowledge stacks and MCP together? Yes. In a single conversation, the AI can reference your indexed documents and call MCP tools for live data. The model decides which source to consult based on your question.

How does model comparison work with MCP? When comparing models side by side, all models have access to the same MCP tools. This lets you evaluate how different models use external data differently.

Does Msty work offline? With Ollama models — yes. Knowledge stacks work locally. MCP servers require internet connectivity.

Is Msty free? Msty has a free tier with basic features. Premium plans unlock advanced RAG, unlimited knowledge stacks, and priority support.