Announcing New Concurrency Features →

Announcing the Hatchet CLI

February 26th, 2026

TL;DR: we just shipped the Hatchet CLI, which you can install via:

curl -fsSL https://install.hatchet.run/install.sh | bash

This lets you interact with Hatchet from your terminal, and is particularly useful for coding agents like Claude Code and Cursor. You can find the full documentation for the CLI here.

Skills and Docs MCP

A frequent ask from Hatchet users has been better support for coding agents, like Claude Code and Cursor, and in particular support for an MCP integration or skills. The new CLI now comes bundled with a handful of skills that early users have found particularly useful, like debugging or replaying runs. This can be installed into your project via:

hatchet skills install

You can also install an MCP which provides helpful context for coding agents, like how to use the SDKs to write tasks and workflows. This can be installed via:

hatchet docs install cursor
hatchet docs install claude-code

For more information, see here: https://docs.hatchet.run/home/coding-agents

Listing, replaying and debugging runs

The CLI provides a bunch of new commands for interacting with runs. For example, you can list all recent failed runs with:

hatchet runs list --since 24h --status FAILED

You can also replay a run with:

hatchet runs replay 8ff4f149-099e-4c16-a8d1-0535f8c79b83

Or a batch of runs based on filters:

hatchet runs replay --since 1h --status FAILED

There much more, like listing logs, child runs, cancellations, and status checks. Run hatchet runs -h for the full set of commands.

Local Development

The CLI comes with functionality for running a worker locally with hot reload, as well as triggering tasks. To run a worker locally, you can run:

hatchet worker dev

This will start a local worker which automatically connects to your Hatchet account, and reloads whenever you make changes to your code. See docs for more details: https://docs.hatchet.run/cli/running-workers-locally.

Hatchet TUI

We’re also shipping a full terminal user interface (TUI) for Hatchet, which you can access by running:

hatchet tui

This TUI provides a way to visualize your tasks and workflow runs from the terminal, without needing to switch over to the web UI. It’s particularly useful for users who are developing Hatchet tasks and durable workflows in their IDE:

… and more!

The CLI also comes with additional commands for managing your Hatchet account and resources, like:

Subscribe to Hatchet product updates