naetive

Your first ten minutes

Last reviewed: 16 Sep 2026

This page takes you from signing in to an agent that already knows your project. No coding; ten minutes if you have Claude Code, two if you connect a chat agent instead.

This is the how-to. For the idea behind it, read Working With Naetive (the tab next to Docs).

What you will have at the end

  • One agent connected to Naetive, acting as you
  • A project with its first understanding: what it is, how it is built, what shaped it
  • The next agent you open starting from that understanding instead of from zero

Step 1: Connect an agent

Open Connect in the sidebar and pick the agent you use.

  • Claude Code: click Install. A small launcher downloads; double-click it and a tab opens asking you to approve this computer. Click Approve and the installer finishes on its own. Your key is never shown, pasted or typed. It sets up Claude Code (installing it if you do not have it), the Naetive connection, the activity hooks, the local console and the tunnel.
  • ChatGPT, Claude on the web, Claude Desktop, Codex: open that agent's row and follow the two or three steps shown. These connect in a minute and need nothing installed.

The row turns to "set up" the first time that kind of agent says hello. See How to connect for which path suits which agent.

Step 2: Point the agent at your project

Naetive does not read your code by itself. An agent does, and writes what it understands.

  • You have code already: open your agent in the repository and say "Map this project in Naetive." It reads the code, writes an orientation and the first entries, and creates a couple of jobs from what it finds unfinished.
  • You have an idea and no code yet: say "Set this up in Naetive: one line about the idea." The agent creates the project and starts its understanding from what you tell it.

Connect's third step walks you through this the first time.

Step 3: Watch it land

  • Status is the front door: one word for how the project's understanding is doing, and four cells for understanding, agents, work and code.
  • Understanding is the map: every entry the project knows, by topic, with the file it cites.
  • Work is what your agents are doing: who is here, what is in motion, what needs you.

The next agent you open, on any of your tools, starts from this. That is the loop.

Read next

  • How to connect: which agent, which path, what each gives you
  • Start a project with your AI: the one-time instruction that makes every chat start grounded
  • The folder: carry the understanding in your repo, so a fresh clone inherits it