quantustik

Build your own market assistant on Claude

A working setup in under two minutes, no signup required to start.

Quantustik is a risk-first market-intelligence layer for humans and AI agents, delivering calibrated probabilistic forecasts, transparent track records, and machine-readable decision support through REST and MCP. This page connects that layer to Claude, so you can ask in plain English and get the same calibrated forecasts, signals, and risk plans the dashboard shows.

Disclaimer. Everything below is educational: how to point an AI client at Quantustik's own data and read the result. Nothing here is investment advice, a solicitation to buy or sell any security, or a personalized recommendation. Signals and forecasts are probabilistic model outputs with explicit uncertainty, not promises. Write your own prompts to ask for data, analysis, and explanation, never "tell me what to buy." Full text: Disclaimer & Terms of Use.

1. Add the Quantustik connector

Claude Desktop or claude.ai: Settings → Connectors → Add custom connector, then paste https://quantustik.com/mcp.

Claude Code: run this in a terminal.

claude mcp add --transport http quantustik https://quantustik.com/mcp

No API key is required to start; anonymous use shares a generous hourly cap. An optional free key at quantustik.com/developers raises that cap if you plan to ask a lot of questions.

2. Confirm the connection

Paste this into a new chat once the connector shows as connected.

List everything the Quantustik MCP server can do.

Claude should call the server's list_capabilities tool and describe roughly two dozen tools, three guided prompts, and four reference resources back to you.

3. Ask a first question

A minimal end-to-end example: paste the prompt, Claude calls the tool, you get a real answer back.

Using the Quantustik get_signal tool, what is the current signal for MSFT? Give me the entry, stop, and take-profit levels, and explain in plain language what is driving the call and what would invalidate it.
Illustrative example. Claude calls get_signal("MSFT") and might come back with something like: verdict WAIT, entry $412.50, stop $398.10, take-profit $438.00, with a note that the market conditions read is Neutral and the dominant factor is a flattening momentum signal. The exact numbers change every trading day; run the prompt yourself for the current reading.

4. Turn it into a standing assistant

Create a Claude Project (or a persistent chat) named something like “My market assistant” and paste this into its custom instructions, so every new conversation starts from the same rules.

You have access to the Quantustik MCP server. When I ask about a stock, use get_signal and get_forecast for it rather than your own training data. When I ask about several stocks at once, use get_signals_batch. Always state the confidence band or the invalidation level alongside any verdict. Never tell me what to buy; report what the tools return and explain the reasoning.

5. Ask about your own tickers

Check AAPL, NVDA, and TSLA with get_signal. For each, tell me its verdict and timing labels and summarize the risk in one sentence. Report what the tools return; do not tell me what to buy.

Swap in whatever tickers you actually hold or watch. Your Quantustik watchlist works as a source for the list if you keep one there.

More ways to run this

Disclaimer. Everything below is educational: how to point an AI client at Quantustik's own data and read the result. Nothing here is investment advice, a solicitation to buy or sell any security, or a personalized recommendation. Signals and forecasts are probabilistic model outputs with explicit uncertainty, not promises. Write your own prompts to ask for data, analysis, and explanation, never "tell me what to buy." Full text: Disclaimer & Terms of Use.