API Overview

SendHub.ai is built on Supabase — your API is a fully typed PostgreSQL REST API with real-time subscriptions and row-level security out of the box.

You can interact with your SendHub data in two ways:

Method	Best for
Supabase Client	Apps, integrations, dashboards — full type safety, realtime, any language
MCP Server	Connecting AI agents (Claude, Codex, Gemini, Cursor)

Install the Supabase Client

npm install @supabase/supabase-js

Initialize

Your API Secret Key is available in the SendHub dashboard under API Keys.

import { createClient } from '@supabase/supabase-js'

const API_SECRET_KEY = 'your-api-secret-key' // from Dashboard → API Keys

const sendhub = createClient(
  'https://noddeqgxbyujhpqxezbg.supabase.co',
  'sb_publishable_YVA7_HVuhhguR-OVA-LZPg_ecJF2CY2',
  { global: { headers: {
    Authorization: `Bearer ${API_SECRET_KEY}`
  }}}
)

// All queries are scoped to your org automatically
const { data } = await sendhub
  .from('conversations')
  .select('id, status, last_message_at, contacts(name, phone)')
  .eq('status', 'ai_handling')
  .order('last_message_at', { ascending: false })

Enterprise and self-hosted customers use their own Supabase project. You’ll need all three credentials from your project settings.

import { createClient } from '@supabase/supabase-js'

// All three values from your Supabase project
const PROJECT_URL = 'your-project-url'
const ANON_KEY = 'your-anon-key'
const API_SECRET_KEY = 'your-api-secret-key'

const sendhub = createClient(
  PROJECT_URL, ANON_KEY,
  { global: { headers: {
    Authorization: `Bearer ${API_SECRET_KEY}`
  }}}
)

const { data } = await sendhub
  .from('conversations')
  .select('id, status, last_message_at, contacts(name, phone)')
  .eq('status', 'ai_handling')
  .order('last_message_at', { ascending: false })

TypeScript Types (Optional)

If you’re using TypeScript, you can generate types for full autocomplete and type safety:

npx supabase gen types --lang typescript \
  --project-id noddeqgxbyujhpqxezbg > types.ts

Schema changes are announced in the SendHub dashboard. When the schema changes, re-run this command to keep your types in sync.

Data Model

Organization (billing, plan, team)
  └── Business / Workspace (AI config, channels, contacts)
        ├── Channel (WhatsApp number — business or personal)
        ├── Contact (WhatsApp user)
        └── Conversation (thread between contact + channel)
              └── Message (individual message with delivery tracking)

Concept	Table	Description
Organization	`organizations`	Top-level tenant. Owns billing, plan limits, team members.
Workspace	`businesses`	A workspace within an org. Has its own AI config, channels, contacts.
Channel	`channels`	A connected WhatsApp number (business API or personal bridge).
Contact	`contacts`	A WhatsApp user who has messaged your channels.
Conversation	`conversations`	An ongoing thread on a specific channel with a specific contact.
Message	`messages`	Individual message — inbound or outbound, with delivery status and AI metadata.
Team Member	`team_members`	A user with access to the org (owner, admin, or agent).
API Secret Key	`api_keys`	Programmatic access keys with scoped permissions.

Row-Level Security

Every table has RLS policies. Queries are automatically scoped to your organization — you can only access data where you’re a team member. No manual org filtering needed.

Quick Examples

// List all conversations with AI handling
const { data } = await sendhub
  .from('conversations')
  .select('*, contacts(name, phone), channels(label)')
  .eq('status', 'ai_handling')

// Send a message
await sendhub.from('messages').insert({
  conversation_id: 'uuid',
  business_id: 'uuid',
  org_id: 'uuid',
  direction: 'outbound',
  sender_type: 'agent',
  content: 'Your order has shipped!'
})

// Subscribe to new messages in real-time
sendhub
  .channel('inbox')
  .on('postgres_changes', {
    event: 'INSERT',
    schema: 'public',
    table: 'messages',
  }, (payload) => console.log('New:', payload.new))
  .subscribe()

Next Steps

Authentication — API secret key setup for server-side use
Client-Side API — Browser apps with user auth, no secret key needed
Database Schema — Full table reference
MCP Server — Connect AI agents to your data