Using Coding Agents

Hatchet is designed to work well with AI coding agents. This page covers how to give your agent tools for operating Hatchet, step-by-step skills for common CLI operations, and access to Hatchet documentation.

Agent Skills

Agent skills are reference documents that teach AI coding agents how to use the Hatchet CLI: triggering workflows, starting workers, debugging runs, and more.

Run the following command in your project root to install the skill package:

hatchet skills install

This creates a skills/hatchet-cli/ directory with step-by-step reference files and appends a section to your project's AGENTS.md (and CLAUDE.md) pointing agents to the right file for each task.

Install to a custom directory:

hatchet skills install --dir ./my-project

After installation, commit the skills/ directory and AGENTS.md to version control so all agents working in the repo benefit automatically.

Available references

Local Development with Embedded Mode

For local development, an agent does not need an API token or Docker. Embedded mode runs a full Hatchet engine locally, so an agent can start a worker, trigger tasks, and inspect results without any account setup.

Each SDK has an embedded entry point: Python uses Hatchet.from_embedded(), TypeScript uses HatchetEmbeddedClient.init() from @hatchet-dev/typescript-sdk/v1/embedded, and Go uses hatchet.WithEmbedded() with a blank import of github.com/hatchet-dev/hatchet-embedded:

hatchet = Hatchet.from_embedded()

hatchet = Hatchet.from_embedded()

See Embedded Mode for configuration options, stopping the engine, and serving the dashboard.

MCP Server

The Hatchet CLI includes a local MCP (Model Context Protocol) server that lets coding agents operate against a running Hatchet deployment: trigger workflow runs, inspect run status and events, list workers, replay runs, check engine status, and submit product feedback.

Add it to your agents with one command:

hatchet mcp install

Use --target for non-interactive setup:

hatchet mcp install --target claude-code,cursor

Access is default-deny, you must grant access to CLI profiles during the installation process or with hatchet mcp auth after installation. The only exception is embedded instance, which is always available without an explicit grant.

See the MCP server reference for all targets, scopes, and the grants model.

llms.txt

For any AI tool that supports llms.txt, Hatchet docs are available at:

Every documentation page also includes a <link rel="alternate" type="text/markdown"> header pointing to its markdown version, and a "View as Markdown" link at the top of the page.