What changed
The Pathiel project introduces an AI trading agent that connects to the Hyperliquid perpetual futures exchange. This agent functions as an MCP (Multi-Client Protocol) server and a Claude Code plugin, enabling AI models like Claude to interact with market data and execute trades. It supports both research-only modes, which only read market data and account information without placing orders, and full trading modes. For trading, Pathiel requires a Hyperliquid API wallet and automatically sizes trades based on a user-defined risk profile, defaulting to 3x leverage and 1% of equity per trade. Each order includes an exact stop-loss and take-profit, forming a bracket order. The agent provides desk briefs for various markets, including crude oil, metals, and the S&P 500, offering AI models written calls to inform their trading decisions. Installation is streamlined, with options for direct integration into Claude Code via plugin marketplace commands or through shell commands and Docker containers for broader MCP client compatibility. The project is released under version v0.9.0 and is written in Python.
Why it matters for builders
Pathiel provides AI developers with a concrete example of how to integrate AI agents into high-frequency trading environments. It demonstrates a practical application of AI for financial markets, showcasing how complex decision-making and order execution can be automated. The project's open-source nature and clear integration paths via MCP and SDKs allow builders to experiment with, adapt, and extend its capabilities for their own trading bots or AI-driven financial tools.
Practical impact
Builders can leverage Pathiel to develop AI trading strategies that execute trades with precise risk controls directly from their AI models. The project's integration with Claude Code and the Claude Agent SDK means that developers can start experimenting with trading prompts and see how the agent responds. For those with a Hyperliquid account, the setup process is designed to be quick, allowing for the testing of real-money trades with a default risk configuration. The availability of desk briefs also offers a structured way to feed market analysis into AI decision-making processes. Developers can explore the provided prompts to understand the agent's capabilities, from retrieving market data and account information to sizing and placing bracket orders.
Caveats and source limits
This analysis is based on the provided GitHub repository information and excerpt. Specific details regarding independent performance benchmarks, real-world trading results, or advanced customization options beyond the described setup are not detailed. The excerpt mentions a "fresh release" and version v0.9.0, indicating that the project may still be under active development and subject to changes. While it supports trading on Hyperliquid mainnet, users are advised to practice on testnet first. The exact parameters for the "3x / 1% risk by default" can be configured, but specific default values beyond this description are not enumerated. The excerpt does not provide pricing information, as the project is open-source and free.
Sources
Claim check: 7/7 supported claims - 7 evidence links - 100% avg confidence
- Pathiel is an AI trading agent for Hyperliquid that trades perpetual futures.supported - github.com
- The agent integrates with Claude Code and other MCP clients.supported - github.com
- Pathiel places orders with an exact stop and target for every trade.supported - github.com
- The default risk profile is 3x leverage and 1% of equity per trade.supported - github.com
- Pathiel provides desk briefs for oil, metals, and S&P 500.supported - github.com
- The project is free and open-source.supported - github.com
- The latest release is v0.9.0.supported - github.com
Caveats
- Single-source caution: verify critical details at the linked source.
Radar score 87/100 - how it was calculated
- Reliability 82: GitHub metadata supports source trust
- Freshness 100: Fresh GitHub release date
- Novelty 77: Fresh GitHub release
- Technical 85: Repository technical metadata
- Developer 96: Developer tooling signals
- Ecosystem 72: Fresh GitHub release
- Confidence 96: Claims have reliable evidence
Discussion
Loading comments...