edhil is a command-line tool that turns a written plan into working, verified code. You describe a project as an epic: a list of small work items, each with clear steps, a definition of done and the commands that prove it is finished. edhil imports the epic into your GitLab project, then hands each work item to an AI coding agent in turn.

After the agent finishes a work item, edhil runs that item’s own verification commands. If something fails, the agent is shown the failure and gets a chance to fix it. Every work item that passes ends in its own commit, so the history is clean and easy to review. If the agent gets stuck, the run pauses and waits for you instead of carrying on. A full journal of every step, command and result is kept, so you can always see what happened and why.

edhil works with the agent and the tools you already use, and it keeps you in control: you write and approve the plan, you set which commands may run and how long they may take, and you decide what to do when something goes wrong.

Documentation topics

Getting started

Using edhil

Commands