Getting started with nectar

Nectar doesn't need its own install step or its own setup. It comes with the rest of the Apiary, and once it's running, you use it the same way you already talk to your assistant.

#How do I install nectar?

You don't install it on its own. One command installs the whole Apiary stack, and nectar comes with it:

curl -fsSL https://get.theapiary.sh | sh

That's it. Nectar's daemon starts on 127.0.0.1:3854 and is supervised by doctor along with the rest of the stack. There's no separate download and no separate service to manage.

#Do I need to configure anything?

No. There's nothing to set up. Nectar reads your project, writes its descriptions, and stays out of your way. You don't add comments, naming conventions, or any kind of marker to your code. You keep writing and organizing code exactly as you already do.

#How do I use it?

The easiest way is to just ask your assistant a plain question about your code:

"Where do we handle user authentication?"

Your assistant pulls in nectar's descriptions through honeycomb's shared recall automatically. You don't have to invoke anything by name for this to work.

If you want to search directly yourself, the nectar search command does the same thing from the terminal:

nectar search "where do we handle user authentication"

You'll get back the files that actually do the work, even ones that aren't named anything like "auth."

#What happens the first time nectar sees my project?

The first read-through of your codebase is called brooding. It's the one-time step where nectar reads each file and writes its first description. You can preview what that will cost before it spends anything:

nectar brood --dry-run

This makes no model calls and writes nothing. As a rough guide, about 200 files runs around thirty cents, 2,000 files around three dollars, and 10,000 files around fifteen dollars.

When you're ready, run the real scan:

nectar brood

After that first scan, there's almost nothing to do day to day. Nectar watches your source and only re-describes a file when it changed in a way that matters, and only after you've paused editing.

#How do my teammates get this without paying the cost again?

Commit the map nectar writes:

git add .honeycomb/nectars.json
git commit -m "add nectar understanding"

Anyone who clones the repo after that inherits every description instantly, for free, with no re-scan needed. One person pays the one-time cost. Everyone else just gets it.

#Common questions

Do I install nectar separately? No. It installs with the rest of the Apiary in one command and runs as its own supervised daemon.

Is there configuration involved? No. Nothing to set up. You ask questions and it works.

What's the fastest way to try it? Run nectar brood --dry-run to see the cost, then nectar brood to scan, then ask your assistant a real question about your code.

#Next steps

• How nectar works
• Nectar reference
• Nectar FAQ