nectar
Find code by meaning, not by file name. Ask your agent where you handle logins and it should hand you the right files, even when none of them are named login.ts. That is the job nectar does.
meaning, not matching
nectar reads each file and writes down what it does in plain language, something like refreshes login tokens on each authenticated request. It stores that alongside the file and keeps it current as files change, so your agent matches on meaning instead of names, and finds the right file even when it is poorly named or buried three folders deep.
what it does
meaning, not matching
Every file gets a short plain-language description of what it actually does, searchable by intent.
quietly current
Descriptions are written once and kept up to date as files change, move, or get renamed. You do nothing differently.
fewer dead ends
Your agent points at the file that does the work, not the file that happens to share a name with your question.
team-ready
Once one person builds the understanding, the rest of the team gets it for free when they pull the project.
the specs
- daemon
- 127.0.0.1:3854
- surfaces
- auto via honeycomb recall, plus nectar search
- data
- hive graph + portable registry
- portability
- understanding ships with the project
- embeddings
- provider switching supported
- identity
- minted descriptions, trustworthy on a team
common questions
Is nectar a full-text code search engine?
No. It is a memory layer. You can query it with nectar search, but the same recall also surfaces automatically through your assistant via honeycomb, with no search box at all.
Does it replace my editor or my agent?
No. It does not write code and it does not replace the assistant you already use. It makes that assistant smarter about your project.
Does it read every line of my code?
No. nectar reads enough of each file to describe it accurately. It does not memorize your source line by line.
How does the team get it?
The understanding is portable. Once it is built, it ships with the project, so a teammate gets it for free on download.
give your agent an index that read every chapter
One command sets up the whole apiary, nectar included. Then ask your agent the same questions and get the right files back.
Windows (PowerShell): irm https://get.theapiary.sh/install.ps1 | iex
get nectar