Two commands put Stele on your machine. One more, inside the agent you already use, wires it into your project. After that you keep prompting the way you always have and the record fills in around you.
Three steps, about two minutes.
One line, on macOS, Linux, or Windows. It installs a single binary, finds the coding agents you already have, and asks which ones to set Stele up for. You can run it again later to add more.
curl -fsSL https://app.stele-ai.dev/install.sh | bashNothing to configure, and no runtime to install first.
Your browser opens, you approve, and you are back in the terminal. Use GitHub, Google, or an email link. There is no key to paste and no second account to create.
stele account loginOpen the project in your agent and run this once. Stele reads the codebase and any docs it finds, builds the first record, and shows you what it is about to do before it does it.
/stele:startIn Claude Code and Cursor this is a slash command. In other agents it is a skill or a command with the same name. If the agent says it does not know the command, restart the session so it picks up what the installer added.
That is the setup. From here you prompt your agent as usual. Before it answers, Stele hands it the decisions, lessons, and risks that bear on what you asked; as the work happens, it writes new knowledge and tasks back.
Restarting the agent fixes most of it. If it still does not, the walkthrough covers each agent's differences step by step.