For the App Intents and agent-native era

Build the personal agents people will trust.

A Personal Agent Platform for building agents that understand users through consented context.

Find subscriptions I no longer use.

WWDC watchApple Extensions SDK support is coming soon.
How it works
Consent-aware agents, built for many surfaces.
Hussh
User asks
Find subscriptions I no longer use.
Nav
Confirms what the user approves and keeps a record.
Agent acts
Runs where it fits best: local, BYOC, cloud, or Python.
Surfaces
App Intents, MCP, iMessage, BYOC.

From idea to consent-aware agent.

A developer should not need to invent consent, context, runtime packaging, and review flow from scratch.

01
Describe an agent
02
Choose a runtime
03
Declare context
04
Test consent behavior
05
Export or register

Aligned with where Apple is taking app actions.

WWDC is making one thing clear: apps are becoming action surfaces for system intelligence. App Intents, Siri, Shortcuts, Spotlight, and Foundation Models point toward agents that can act with context, not just sit inside one app.

App Intents

Expose agent actions as discoverable system capabilities.

App Entities

Make agent-owned content easier for system intelligence to find and pass around.

Shortcuts and Spotlight

Let people invoke useful agent actions outside the app.

Foundation Models

Use private on-device reasoning where the runtime belongs on Apple devices.

Build from your coding tool with Hussh MCP.

Connect Codex, Cursor, Claude Code, Windsurf, Antigravity, or any MCP-capable tool. Hussh exposes the agent build path as tools your coding assistant can call.

Scaffold

Generate the manifest, README, and starter agent files.

Scope

Add Nav-readable context reasons and fields.

Validate

Check the agent contract before it runs.

Test

Run sandbox personas and consent behavior.

Package

Create the downloadable agent bundle.

Register

Send the package for sandbox review.

What Hussh adds to the agent stack.

App platforms expose actions. Models generate output. Hussh provides the consent, context, runtime, sandbox, and proof layer that makes personal agents safe to build.

Consent-aware context
Nav approval flow
BYOC and BYOK runtimes
MCP builder tools
Sandbox personas
Review-ready packages

Run where the agent belongs.

A personal agent platform should not force one model, one framework, or one runtime. Hussh lets you build for Apple devices, BYOC services, sandbox previews, and MCP-connected tools.

Apple devices and iMessage extensions
MLX and on-device models
OpenClaw and BYOC services
Cloud models with BYOK
Google ADK and Python agents
MCP-capable coding tools
Join Program

Join with an agent idea. Build from a clear platform path.

Signup creates your builder workspace and carries your first agent idea into the MCP build flow.

Your workspace includes
MCP builder
SDK downloads
BYOC runtime guide
BYOK sandbox setup
Agent package download
Sandbox registration