Keel — Hyperliquid Trading Agent
Backtest typed Hyperliquid strategies in your coding agent and promote the same logic to live automation with deterministic parity.
Overview
Keel Trade is a Finance-phase MCP server that lets coding agents build, backtest, and automate Hyperliquid trading strategies with typed, deterministic live-parity-oriented workflows.
What is this MCP server?
- Typed, deterministic strategy definitions aimed at matching backtest and live behavior
- Hyperliquid-focused build, backtest, and automate workflow inside an MCP agent
- stdio MCP via PyPI (`keel-trade mcp serve`) or installable MCPB bundle v0.5.6
- Documented live parity goal so agents can reason about strategy changes consistently
- Server schema version 0.5.6 with PyPI identifier `keel-trade` and optional MCPB release artifact
- stdio transport via positional args `mcp` and `serve`
- Repository: github.com/keel-trade/keel-trade
What problem does it solve?
Solo builders waste cycles when backtest scripts diverge from live Hyperliquid bots and agents cannot safely iterate strategies in one consistent toolchain.
Who is it for?
Indie quant-curious builders already comfortable with Hyperliquid who want agent-driven backtests before turning on automation.
Skip if: Complete beginners seeking passive investing advice or anyone who will not manage exchange keys, risk limits, and live-trading liability.
What do I get? / Deliverables
After install, your agent can prototype and backtest Hyperliquid strategies through MCP and follow the same typed definitions toward automated live trading.
- Agent-callable MCP tools for Hyperliquid strategy build, backtest, and automation flows
- Reusable typed strategy artifacts intended to align simulation with live execution
- Documented stdio or MCPB-based server registration for local agent sessions
Recommended MCP Servers
Journey fit
Strategy proof and simulation belong in Validate before you risk capital; Keel’s backtest-and-automate loop is the gate between idea and production trading. Prototype is the canonical shelf for running historical simulations and iterating strategy code before live deployment on Hyperliquid.
How it compares
Hyperliquid trading MCP integration, not a generic portfolio spreadsheet skill or discretionary signal newsletter.
Common Questions / FAQ
Who is io.github.keel-trade/keel-trade for?
Developers and solo builders automating Hyperliquid strategies who want MCP-backed backtesting and live automation from the same typed strategy code.
When should I use io.github.keel-trade/keel-trade?
Use it while validating a strategy in simulation on Hyperliquid and when you are ready to graduate tested logic toward automated live execution via your agent.
How do I add io.github.keel-trade/keel-trade to my agent?
Add the MCP server to your host config with stdio transport running the PyPI package (`keel-trade mcp serve` at v0.5.6) or install the MCPB bundle from the GitHub release, then wire Hyperliquid credentials per Keel docs.