What if all your AI coding tools finally shared one memory?
Your assistants are brilliant and forgetful, and each one hoards what little it learns. The Apiary is a stack of small, sharp programs that give all of them one shared memory on Deeplake, on hardware you control, each fixing one stubborn problem. Learn something once, and every agent recalls it everywhere.
curl -fsSL https://get.theapiary.sh | sh 
Why do you re-explain your own project to your AI every morning?
You taught your assistant your stack and the fix that finally worked at midnight, then closed the window and it forgot all of it. Honeycomb is the memory it never had: install once and it hands the right notes back to any assistant, on any machine. Learn something once, recall it everywhere.
I told Cursor yesterday. Why does Claude Code have no idea today?
Their memory dies with the window, and they do not talk to each other. Honeycomb sits underneath all of them (Claude Code, Cursor, and Codex today) and gives them one shared brain, so a note written in one is recalled by another, even on a different laptop.
Do I have to babysit it?
No. Install opens the Hive dashboard, where you connect your assistants in a couple of clicks, and from then on it captures the useful moments on its own and briefs every new session. Your agent starts smart instead of blank.
How does it find the right memory when I forget the exact words?
Your memory lives on Deeplake, which searches by meaning and not just keywords, so it surfaces the note you needed even when you would never have guessed the term. It also tidies itself over time, so the more you use it the sharper it gets, not the noisier.
Is my code safe?
Yes. Your memory sits on Deeplake, reached only by the local helper on your machine and never by the assistant directly, with secrets never shown to the agent. Working on something sensitive? Flip to read-only and nothing new is written, while recall still works.
Why does your agent open the wrong file when you ask a simple question?
Ask where you handle logins and your agent hunts for login.ts, misses the real file buried three folders deep, and hands you a confident answer about the wrong code. Nectar gives every file a plain-language description of what it actually does, so your agent finds code by meaning and stops guessing from names.
Why can't my agent find code that isn't named after what it does?
Because it searches names and keywords, and real codebases hide login logic in a file called session-refresh.ts. Nectar reads each file and writes down what it does, so a search for logins lands on the right one even when the name never says so.
Do I have to change how I work?
No. Nectar runs quietly and keeps its descriptions current as files change, and the better answers surface right inside your assistant through Honeycomb's shared memory. You ask the same questions and get the right files back.
How does it match meaning instead of keywords?
The descriptions live on Deeplake, which searches by meaning, so 'anything about logins' finds the right file even when you would never have guessed its name or where it sits. It is the book index that actually read every chapter.
Will it keep up as the code moves?
Yes. Nectar re-reads files as they change, so a renamed or relocated file still gets found by what it does. Poorly named, buried, or freshly moved, your agent still lands on it.
What happens to your agent's memory when a daemon dies at 2am?
Your memory stack runs on local daemons, and when one dies quietly overnight you pay the next morning: a session that forgot everything and twenty minutes re-explaining a codebase your agent knew yesterday. Doctor watches the whole stack, heals what breaks, and has your memory back before your next prompt, usually without you noticing.
A daemon crashed. Do I have to fix it?
No. Doctor probes every service every 30 seconds and climbs a repair ladder: restart, reinstall after repeated fails, remove a conflicting package, back off, and stop the instant health returns. Kill one on purpose and it is typically back within one probe.
Will it nag me?
No. A healthy check is a single debug line, so Doctor stays silent on the happy path. It only speaks up when the ladder runs out, and then it writes a plain do-this-next report on a local status page.
Could an auto-update brick my setup?
No. Doctor updates the memory daemon only behind a blessed-release gate, verifies health afterward, and rolls a bad update back automatically. A broken release cannot spread itself to your machine.
Can it touch my credentials?
Never. Doctor has no code that can read or delete your credentials file. If it suspects a credential problem it tells you and stops, full stop.
Which local port was the dashboard on again?
The Apiary runs several services on your machine, each on its own port, and hunting across loopback tabs is nobody's idea of a dashboard. Hive is the front door: one always-on address at 127.0.0.1:3853 that serves the whole thing. Bookmark it once and you are done.
Why one address instead of a port per service?
Because port-hunting is not your job. Hive serves memories, projects, the graphs, sync, logs, ROI, and settings from a single URL, with a live health rail on every page so you always know the state of the fleet.
What do I see if I open it mid-boot?
An honest readiness screen, one bee per service, that dismisses itself the moment the fleet is ready. You never get a broken page or a false first-time-setup screen just because something was still waking up.
A panel says unreachable. Is my data gone?
No. It means that one service is not answering right now; Doctor restarts it and the panel recovers on its own. The rest of the dashboard keeps working the whole time.
Is it exposed to the network?
No. Hive binds to 127.0.0.1 only, so nothing off your machine can reach it, and it passes your session straight through to the services that own your data while storing nothing itself.
Your agents are spreading across machines. How do you keep them one identity, not a mess?
The Apiary is clean on one laptop. The moment you add a second machine, a VPS, or throwaway workers, you need to know which agents are alive and who is allowed to join, without handing your memory to the cloud. Queen coordinates all of that and never reads your memory content. Coming soon.
Do I have to open two laptops to add a device?
No. Approve the new device in the cloud and an existing trusted machine finishes the cryptographic handoff next time it is online. A headless VPS joins with a short-lived token whose only power is to let it in, it cannot read or decrypt anything.
Can I see all my agents in one place?
Yes. A read-only fleet view shows every agent with a derived health state, so an idle-but-fine daemon and a crashed one stop looking identical. It is scoped to your own fleet, nothing more.
Does the cloud hold my credentials or memory?
No. Your orchestrator holds the Deeplake credential; the cloud coordinates identity and presence and stores only encrypted blobs it cannot open. No prompts, no session text, no plaintext keys.
What if I lose a device?
Revoke it and rotate your credential, two clear steps, and it is cut off. Lose every trusted device and the answer is a written re-link path, never a hidden backdoor.
One memory, a whole hive around it.
Install the stack with one command. Bookmark one dashboard. Learn something once, and every agent recalls it everywhere.