the apiary
Download
honeycomb docs
GitHub Community latest

What is Nectar?

The plain-language introduction to Nectar: what it is, the problem it solves, and who it is for. Start here if you are new. Written for people who use Nectar, not people who build it.

What is Nectar?

The plain-language introduction to Nectar: what it is, the problem it solves, and who it is for. Start here if you are new. Written for people who use Nectar, not people who build it.

Related:

#The problem: your agent looks for the wrong files

Modern AI coding assistants are good at reading code and bad at finding the right code to read. Ask "where do we handle logins" and the assistant hunts for a file named login.ts or auth.ts. In a real codebase, that logic often lives in a file like session-refresh.ts, buried three folders deep, that no name-based search would guess. So the assistant reads the wrong files, gives a confident answer that is wrong, and you end up searching yourself.

#The idea: describe every file in plain language

Nectar reads each file in your project once and writes a short plain-language description of what it does, for example "refreshes login tokens on each authenticated request." It stores that description and keeps it current as your code changes. When your assistant later searches for anything about logins, it searches those descriptions, not just file names, and finds the right files by their purpose.

Think of it as the index at the back of a book, but one that has actually read every chapter and lists what each chapter is about, not just which words appear on the page.

#What you actually get

  • Search that matches meaning. The assistant finds the file that does the work, whatever it is named.
  • Understanding that survives refactors. Each file gets a stable identity, so its description follows it through renames and moves. Reorganizing your folders does not reset anything.
  • A low, one-time cost. The first scan of about 2,000 files costs around three dollars. You can preview the exact figure before spending anything.
  • Free sharing. The understanding is committed as one file, so teammates who clone the repo inherit every description instantly and offline.
  • Your source, untouched. The only file Nectar writes is its own map. Not one character of your code is changed.

#What you do differently

Nothing. You write, name, and organize your code exactly as you do today. No special comments, markers, or naming conventions. You ask your assistant the same questions and get noticeably better answers.

#Who it is for

  • Developers and teams using an AI coding assistant who want it to actually find the right code.
  • Anyone working in a codebase they did not write, or one that is messy or poorly organized.
  • Teams that want shared, inherited understanding instead of everyone re-indexing on their own.

#Where it fits

Nectar is one of the five products in the Apiary. Honeycomb holds the memory; Nectar makes your codebase searchable by meaning and feeds those results into Honeycomb's shared recall, so your assistant benefits without a separate search box.

#Next steps