The CLI for Clearly

The command line
for your agent cloud.

Beehaven is the command-line interface to your agent cloud. Connect to a workspace, team, or app; give your agents vector memory; run actions; and drive the canvas — all from the terminal. Software built for agents, not dashboards.

Free to startWorks with Claude & ChatGPTMCP-readyNo walled gardenOne tool, your terminal
$ beehaven connect home # your personal workspace
$ beehaven remember "Refunds: 30 days, minus shipping."
$ beehaven recall "what is our refund window?"
→ Refunds: 30 days, minus shipping.
The workflow

Connect, then call

Every cloud resource — your workspace, a team, an installed app — is addressable by name. Connect to set the target, then call actions against it. Beehaven handles auth and routing; you stay in the terminal.

$ beehaven connect team/acme # a team’s agent cloud
$ beehaven actions # list what this agent can do
$ beehaven call send-digest '{"to":"ops","range":"24h"}'
→ { ok: true, delivered: 1 }
The thesis

Software for agents, not dashboards

Agents are slow and brittle when they have to click through software built for humans. Give them a machine-readable interface instead — a CLI and an MCP surface — and they become first-class users of your cloud.

Memory

remember / recall — give any agent vector memory, retrieved by meaning. The substrate behind your company brain.

Drive the canvas

canvas-invoke reaches into an open Vector Studio tab and runs a registered action — paste a URL, re-trace, export a bundle.

Query state

sql, ls, cat, grep, tree — inspect a workspace like a filesystem, auto-routed to whatever you connected to.

Apps for agents

Install an app and address it by name — beehaven connect <app> — then call its actions. Build software your agents depend on.

How it works

One daemon, every target

A local daemon keeps a secure connection to the Clearly relay and routes every command to the right target. You never manage tokens — sign in once, then address anything by name.

Step
01

Sign in once

One auth, then the daemon holds a secure relay connection. No tokens to copy, rotate, or paste into scripts.

Step
02

Connect a target

connect home, connect team/<id>, or connect <app> — set the target you’re addressing by name.

Step
03

Call anything

Run actions, give agents memory, query state, drive the canvas. Same protocol, one tool, your terminal.

$ beehaven connect home # personal workspace
$ beehaven connect team/<id> # a team
$ beehaven connect <app> # a locally installed app

The mental model: Clearly is the cloud, Beehaven is how you drive it.

Why Beehaven

The CLI for a consumer-grade agent cloud

Not another integration hub. Beehaven is the batteries-included terminal for a maker’s agent cloud — memory, canvas, and apps in one tool.

FeatureBeehavenIntegration hubsHand-rolled scripts
One CLI for every resource
Built-in vector memory
Drive a live web canvas
MCP surface for any agent
Auth + routing handled
FAQ

Frequently asked questions

01What is Beehaven?+
Beehaven is Clearly’s command-line interface and SDK — the way you talk to the Clearly agent cloud from your terminal. You connect to a workspace, team, or app, then call actions, give agents memory, query state, and drive the canvas. It’s a Clearly product, not a separate platform.
02What can I do with Beehaven?+
Beehaven is one command-line tool that connects to your cloud resources (workspaces, teams, apps), runs actions against them, gives agents memory, queries state, and drives the canvas — and manages auth for you. You stay in the terminal; Beehaven handles the relay and routing.
03What does "software for agents" mean?+
AI agents work best with machine-readable interfaces — APIs, MCPs, and CLIs — not visual dashboards built for humans. Beehaven gives agents a first-class interface to Clearly: a CLI they can shell out to and an MCP surface they can call, so an agent can recall company memory, run an action, or export a design without clicking through a UI.
04Does Beehaven work with Claude, ChatGPT, and my own scripts?+
Yes. Beehaven is just a CLI plus an MCP layer, so any agent — Claude, ChatGPT, or your own code — can drive it. That’s how Clearly’s company brain and canvas actions become available to whatever agent you already use.

Drive your agent cloud from the terminal

Connect a workspace, give your agents memory, and run your first action in minutes. Free to start.