Apps

Custom apps, native to every context.

Build any app your organization needs and enable it inside a context — it inherits the context's boundary and sees only what its groups see. Ship it with your AI coding tool, or install one from the store.

Two ways to add an app

Bring your own, or pick one off the shelf.

Build it with your AI coding tool

Everything your AI coding tool needs to scaffold, register and publish a SweetHive app is ready. Describe what you want; it wires the manifest, the scoped API and the surfaces — and ships an app installable into any context.

See the quickstart

Choose from the store

Don't build what already exists. Browse ready-made apps and agents, and enable one inside a context in a click. It goes live scoped — no integration project, no data migration.

Browse the store
Build with your AI coding tool

Create, register, publish.

Point your AI coding tool at the SweetHive app SDK and tell it what the app should do. The instructions, manifest schema and scoped API are ready for it — you review and ship.

Create

Scaffold an app and describe its behavior. Your AI coding tool generates the manifest and the logic against the scoped API.

Register

Register the app to receive an app id and a scoped token. Nothing runs with more reach than the manifest declares.

Publish

Publish it — enable it privately in your own contexts, or submit it to the store for other hives to install.

Example — SDK, CLI and manifest schema are the developer-docs source of truth.

Bounded by the context

An app installed in a context lives inside it.

Apps are not integrations bolted on the side. When you enable one in a context, it inherits that context's boundary: it reads only what its groups can see, from that level down, and it can't reach across the tree without an explicit share. The same scoping law that governs people and agents governs apps — so IT can enable them without opening the whole hive.

Scoped by construction

An app's reach is the context it's installed in ∩ the groups it's granted. Declared in the manifest, enforced in the retrieval layer.

Per-context enablement

Enable an app in one context, a subtree, or the whole hive. Different contexts can run different apps — or the same app, separately scoped.

Audited like everything else

Every action an app takes is attributed and logged with what it read and where it ran. No silent access.

The store

Or just enable one and go.

Browse apps and agents built by SweetHive and its partners, and turn one on inside a context in a click. Installed apps enter scoped — they see exactly what that context's groups see, and nothing else. Build once, list it, and every hive can enable it.

If your organization needs it, you can build it in.

Custom apps and agents, scoped to the contexts they run in — shipped with your AI coding tool or installed from the store.

Get a demo