Quickstart

Goal: a running AI app on Hivemind OS in under 30 minutes.

1. Scaffold

npm create @taskclan/hivemind-app@latest my-app
cd my-app

Pick a template up front with -- --template <name>:

Template	Use it for
`node-starter`	A script: discover the platform, dispatch an intent, recall insights.
`react-chat`	A Vite + React chat UI.
`next-ai-chat`	A full-stack Next.js chat app (server token route + streaming).
`agent-worker`	A server-side agent backend (the learning loop).

2. Configure

cp .env.example .env

Set:

HIVEMIND_BASE_URL — your Hivemind deployment (e.g. https://api.hivemindos.taskclan.com).
HIVEMIND_TOKEN — a user/session token. For local dev, a Supabase access token works.

3. Run

npm install
npm run dev

node-starter prints the platform’s agent count, dispatches identity.whoami, and recalls cross-product insights — proof you’re wired in.

4. Make it yours

Everything is a Hive intent. Dispatch any of them:

import { createPlatformClient } from '@taskclan/platform';
 
const client = createPlatformClient({
  baseUrl: process.env.HIVEMIND_BASE_URL!,
  product: 'my-app',
  getToken: async () => process.env.HIVEMIND_TOKEN!,
});
 
const result = await client.dispatchIntent('your.intent.name', { /* input */ });

Discover what exists: client.listAgents(), client.listSkills(), client.listWorkflows().

Need a capability that doesn’t exist yet? That’s a new intent on the platform — file it, don’t rebuild auth/billing/memory yourself.

Welcome Concepts