Install the Icepick CLI

The Icepick CLI is a command line interface for quickly getting started with Icepick by scaffolding a new agent project:
First, install the Icepick CLI:
npm i -g @hatchet-dev/icepick-cli
Then create a new agent:
icepick create my-agent
Then cd into the agent directory and install dependencies:
cd my-agent && npm install

Generate a Hatchet API token

Icepick requires a Hatchet API token for orchestration. You can generate a Hatchet token by signing up for a free account on Hatchet Cloud or following instructions for self-hosting Hatchet. Add this API token to the .env file in the project:
HATCHET_CLIENT_TOKEN=<my-token>

Set your LLM API key

Add this API token to the .env file in the project:
OPENAI_API_KEY=<your-api-key>

Run your agent

The default icepick agent is a simple deep research agent that can be triggered to perform a deep research task. It uses the OpenAI API to generate responses and the Hatchet API to orchestrate the agent’s execution. For a full walkthrough of this agent, see the deep research example. To start your agent, you can run the following:
npm run dev
To trigger the agent, in a separate terminal you can run:
npm run trigger
This will prompt you for a deep research prompt and then execute the agent.