CLI reference
The honeycomb command-line tool: install, setup, remember, recall, status, dashboard, graph, and the other commands, with what each one does.
The honeycomb command is the single entry point for installing the apiary, wiring it under your coding assistants, and running memory operations like remember and recall. It is a thin client of the local daemon at 127.0.0.1:3850, which is the only process that touches storage.
What is the honeycomb command?
It is the single command-line entry point for the apiary: installing it, wiring it underneath your coding assistants, signing in, and running operations against your memory. It talks to the local daemon at 127.0.0.1:3850, which does the actual work.
How do I save and find a memory from the command line?
Run honeycomb remember followed by a quoted sentence to save a note, and honeycomb recall followed by a quoted question to find it again. Recall matches both exact words and, once warmed up, meaning.
How do I wire up my coding assistants?
Run honeycomb setup to detect and wire every supported assistant on your machine, or honeycomb connect followed by a name to wire just one. Both are safe to run again any time.
The honeycomb command is the single entry point for installing the apiary, wiring it underneath your coding assistants, signing in, and running operations against your memory. It is a thin client of the local daemon (default 127.0.0.1:3850), every command that touches memory, sessions, or the codebase graph sends a request to the daemon, which is the only process that talks to storage. That keeps the command fast to start and keeps all the storage and security logic in one place.
#Command summary
| Command | What it does |
|---|---|
honeycomb install |
Bootstrap entry: bring the daemon up, open the dashboard |
honeycomb setup |
Detect installed assistants, wire their hooks, bring the daemon up |
honeycomb connect <assistant> |
Wire exactly one named assistant |
honeycomb uninstall [<assistant>] |
Remove the apiary's footprint from one assistant, or every detected one |
honeycomb login |
Sign in |
honeycomb status |
Report daemon connectivity, sign-in state, and environment health |
honeycomb update |
Update the command, the daemon, and the assistant bundles |
honeycomb dashboard |
Open the local dashboard |
honeycomb remember "<text>" |
Write a memory |
honeycomb recall "<query>" |
Query memory, both exact words and meaning |
honeycomb agent |
Manage per-agent settings |
honeycomb skill |
Pull, scope, and manage shared skills |
honeycomb secret |
Store, list, and use scoped secrets |
honeycomb graph build / init / pull / diff / history |
Build and inspect the codebase graph |
honeycomb org / honeycomb workspace |
Company and team administration |
honeycomb sessions prune |
Scoped cleanup of captured session history |
honeycomb telemetry |
Inspect operator telemetry |
Two global flags work everywhere: --help (or -h) prints usage, and --version (or -v) prints the installed version.
#What is the honeycomb command?
It is the single command-line entry point for the apiary: installing it, wiring it underneath your coding assistants, signing in, and running operations against your memory. It talks to the local daemon at 127.0.0.1:3850, which does the actual work and is the only process that touches storage. If the daemon is not reachable, a command that needs it reports the failure on one line and exits, rather than printing a stack trace.
#How do I install and set up?
The one-command installer lays down the package, then hands off to honeycomb install, which brings the daemon up, writes an onboarding marker, and opens the dashboard, falling back to printing the local address if it cannot launch a browser. Running it again is always safe.
Once installed, wire your coding assistants:
honeycomb setupThis detects every installed assistant and wires each one, patching its config and registering the memory tools it needs. It is idempotent, so re-running writes nothing where nothing changed and never touches a tool's other settings. To wire a single assistant instead of everything detected:
honeycomb connect <assistant><assistant> is one of the supported names, claude, codex, or cursor today. To reverse the apiary's changes for one assistant, or for every detected assistant when no name is given:
honeycomb uninstall [<assistant>]Uninstall only removes what the apiary added, your other settings and any third-party wiring are left untouched.
#How do I sign in and check health?
honeycomb login
honeycomb statuslogin runs a device-flow sign-in: you approve a code in a browser, and credentials get written locally at file mode 0600. The same flow is also reachable from the dashboard's first time setup button, which is the recommended path for new users. status reports whether the daemon is reachable, whether you are signed in, and which assistants are wired.
#How do I save and find a memory from the command line?
honeycomb remember "we deploy from the release branch, never from main"
honeycomb recall "how do we deploy"remember writes a memory entry through the daemon. recall runs hybrid recall, both exact words and, once semantic search has warmed up, meaning, transparently falling back to word-matching when semantic search is unavailable.
#How do I open the dashboard?
honeycomb dashboardThis opens the local dashboard served by the daemon at the loopback address, showing your memories, connected assistants, shared skills, a codebase map, and overall health.
#How does the codebase graph work from the CLI?
The graph command builds and inspects a live map of your source files, extracted with tree-sitter, no language server and no model involved.
honeycomb graph build # walk the repo and write a fresh snapshot
honeycomb graph init # install a hook that rebuilds the graph after each commit
honeycomb graph pull # fetch a teammate's snapshot for the current commit
honeycomb graph diff <sha1> <sha2> # compare two snapshots
honeycomb graph history # tail the audit log of past builds#What else can I manage from the command line?
honeycomb skill manages shared skills, pulling the latest ones or setting who else you learn from. honeycomb secret stores and lists scoped secrets, an agent can cause a secret to be used without ever reading its value. honeycomb org and honeycomb workspace handle company and team administration, creating, switching, and listing. honeycomb sessions prune cleans up captured trace history you no longer want kept.
#What environment variables change its behavior?
A handful of environment variables are read by the daemon and its hooks rather than passed as flags:
| Variable | Effect |
|---|---|
HONEYCOMB_PORT, HONEYCOMB_HOST, HONEYCOMB_BIND |
Override the daemon's port, host, and bind address |
HONEYCOMB_ORG_ID |
Override the organization selected at login |
HONEYCOMB_CAPTURE=false |
Read-only mode: hooks run but write no trace data |
HONEYCOMB_EMBEDDINGS |
Turn semantic recall off (false or 0) or leave it on |
HONEYCOMB_TELEMETRY=0, DO_NOT_TRACK=1 |
Silence operator telemetry |
#Common questions
#What is the honeycomb command?
It is the single command-line entry point for the apiary: installing it, wiring it underneath your coding assistants, signing in, and running operations against your memory. It talks to the local daemon at 127.0.0.1:3850, which does the actual work.
#How do I save and find a memory from the command line?
Run honeycomb remember followed by a quoted sentence to save a note, and honeycomb recall followed by a quoted question to find it again. Recall matches both exact words and, once warmed up, meaning.
#How do I wire up my coding assistants?
Run honeycomb setup to detect and wire every supported assistant on your machine, or honeycomb connect followed by a name to wire just one. Both are safe to run again any time.