AI agents × Hyperliquid

What an AI agent does on Hyperliquid

Connected to Keel, an AI agent — Claude, ChatGPT, Claude Code, Cursor or Codex — composes a typed strategy, backtests it on real Hyperliquid history and reads the result back to you. 29 tools, one URL, nothing to install.

Ask AI about Keel
Or set it up directly
  1. 1Customize → Connectors → Add custom connector
  2. 2Paste https://mcp.usekeel.io/mcp and choose Connect
  3. 3Sign in or sign up on the Keel page that opens (no account needed beforehand)
Keel MCP endpoint
https://mcp.usekeel.io/mcp

Sign in in the browser when asked. No API keys. Connect → sign in to Keel → Allow

Check it worked

Two ways to tell, and either one settles it — a connection that is already live is not improved by signing in again.

In Claude Keel is listed where you added it (Customize → Connectors → Add custom connector) once the sign-in has finished.

Ask the agent keel_account_status answers with the Keel account the session is signed in as, which nothing but a live connection can return. If the agent says it has no Keel tools, the sign-in did not finish.

Prompt
Use keel_account_status and tell me which Keel account this session is signed in as.

Not yet verified against a Claude version. These steps follow Anthropic's own documentation, read on 2026-09-16.

Ask AI about Keel
Other ways to connect
Every client, step by step
Three framings

Only one of them holds up.

  1. The autonomous bot

    An assistant watching the market and placing orders in real time. It cannot run a backtest in its head, it invents fills it never saw, and every prompt becomes part of the risk.

  2. The one-off script

    Ask for a Python script against the exchange API and you get a file with no portfolio simulation, no fee schedule, no carry accounting and no path to anything tested.

  3. The strategy builder

    What Keel gives the agent: 218 typed components, a DSL the engine validates at compose time, and a backtest on Hyperliquid 15-minute bars and hourly carry with real fees. The agent is good at composition and iteration; the engine does the numbers.

  4. Then the app

    The agent builds, tests and reads. When you want to run a strategy on your account, it hands it to the Keel app and you take it from there.

FAQ

Questions about AI agents on Hyperliquid

Which agents work with Keel?

Claude on web, desktop and phone, ChatGPT, Claude Code, Cursor, Codex and Windsurf — any client that accepts a remote MCP URL. The agents page has the steps for each, with what to expect at every step.

Can an agent find new strategies?

Yes — that is what the tools are for. It searches the component library for signals and regimes, composes a candidate, backtests it, then forks and compares variants until one holds up. The strategy library is there too: ask it to fork a published strategy and change one thing.

Is Hyperliquid the only venue?

Yes. Keel is built around Hyperliquid perps: the price history, the fee schedule, the carry accounting and the component library all assume it, which is what makes the backtest numbers worth reading.

How is this different from an autonomous agent product?

Those keep the assistant in the loop at decision time, so its mistakes are yours in real time. With Keel the assistant works at build time — composing, testing, comparing — and the strategy it produces is a versioned artifact you can read before anything else happens.