honeycomb docs

honeycomb gives your AI coding agents one shared, lasting memory. read what it is, who it's for, and where to start.

honeycomb is the memory layer of the apiary. a local daemon captures what happens in your coding sessions, distills it into notes, and hands the right ones back to any connected assistant, so what one tool learns, they all know.

#What is honeycomb?

Your AI coding assistant is sharp in the moment and forgets everything the second you close the window. Switch from one tool to another and it has never heard of your project. honeycomb fixes that by giving your assistants a single, shared, lasting memory.

A small program (the daemon) runs quietly on your machine. It notices what happens while you work, distills it into clean notes, and hands those notes back to whichever assistant asks, whenever it asks. Learn something once, at midnight, in one tool, and it is there tomorrow morning in a different one.

#Who it is for

Solo developers who are tired of re-explaining their own project to their AI assistant every morning. Install it once, and your decisions, conventions, and fixes stick around.

Teams who want a discovery made by one engineer to reach everyone else automatically. Memory stays scoped to the right company, team, and project, so nothing leaks where it shouldn't, but what's meant to be shared shows up for teammates on their next session without anyone passing a file around.

#What you get

  • Memory that survives. What you figured out yesterday is already summarized and waiting today.
  • Memory that travels. A note written in one assistant is recalled by another. honeycomb plugs in underneath Claude Code, Cursor, and Codex today, with Hermes, pi, and OpenClaw in progress.
  • Skills that spread. A reusable fix or pattern can become a shareable skill that shows up on its own for you and your team.
  • A memory that gets sharper, not noisier. honeycomb periodically tidies itself: merging duplicates, dropping junk, keeping the current version of a fact.
  • A dashboard you can actually read. The hive dashboard shows what's been remembered, how your tools are wired, and the health of the whole system. No database knowledge required.

#Where to start

  • New here? Walk through Getting started to install honeycomb and save your first memory.
  • Want the mental model first? Read How honeycomb works for the capture, distill, recall, and compound loop.
  • Scripting or wiring something up? The CLI and MCP reference covers every command and tool.
  • Have a quick question? Check the FAQ.

#Common questions

What does honeycomb actually store? Clean, distilled notes about your work: decisions, fixes, conventions, gotchas. Not your raw code, though the full raw session is kept too, in case you need to dig back in.

Do I need to know Deeplake or SQL to use it? No. honeycomb hides all of that behind the CLI and the hive dashboard. You install with one command and use plain words: remember and recall.

Is honeycomb part of a bigger product? Yes. honeycomb is the memory of the apiary, the shared toolset built on Deeplake and Hivemind. Other apiary products (like the queen fleet orchestrator) build on the same memory.