packages/examples/supabase/README.md
Deploy AI chat agents as serverless Supabase Edge Functions. These examples show how to run an elizaOS agent as a stateless worker that processes chat messages via HTTP.
All handlers use the full elizaOS runtime with OpenAI as the LLM provider, providing the same capabilities as the AWS Lambda and chat demo examples.
┌──────────────┐ ┌─────────────────┐ ┌────────────────┐
│ Test Client │────▶│ Supabase Edge │────▶│ Edge Function │
│ (curl/deno) │◀────│ Functions │◀────│ (elizaOS) │
└──────────────┘ └─────────────────┘ └────────────────┘
│
▼
┌────────────────┐
│ OpenAI API │
└────────────────┘
| Language | Support Level | Notes |
|---|---|---|
| TypeScript | ✅ Native | Full Deno runtime support |
| Rust | ✅ via WASM | Compile to WebAssembly |
| Python | ❌ Not supported | Supabase Edge Functions use Deno runtime |
Note: Unlike AWS Lambda, Supabase Edge Functions run on the Deno runtime, which only natively supports TypeScript/JavaScript. Python is not supported. Rust can be used via WebAssembly compilation.
Create a .env file in your Supabase project or set secrets:
# Local development
export OPENAI_API_KEY="your-openai-api-key"
# Or set in Supabase Dashboard → Project Settings → Edge Functions → Secrets
supabase secrets set OPENAI_API_KEY=your-openai-api-key
# If starting fresh
supabase init
# Copy edge functions to your project
cp -r examples/supabase/functions/* supabase/functions/
# Start Supabase local development
supabase start
# Serve edge functions locally
supabase functions serve eliza-chat --env-file .env
# Test with curl (in another terminal)
curl -X POST http://localhost:54321/functions/v1/eliza-chat \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_ANON_KEY" \
-d '{"message": "Hello, Eliza!"}'
# Deploy to Supabase
supabase functions deploy eliza-chat
# Set secrets (if not already set)
supabase secrets set OPENAI_API_KEY=your-openai-api-key
# Get your project URL from Supabase Dashboard
curl -X POST https://YOUR_PROJECT.supabase.co/functions/v1/eliza-chat \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_ANON_KEY" \
-d '{"message": "Hello, Eliza!"}'
examples/supabase/
├── README.md # This file
├── functions/
│ ├── eliza-chat/ # TypeScript Edge Function
│ │ ├── index.ts # Main handler
│ │ ├── lib/
│ │ │ ├── runtime.ts # elizaOS runtime manager
│ │ │ └── types.ts # Type definitions
│ │ └── deno.json # Deno configuration
│ └── eliza-chat-wasm/ # Rust WASM Edge Function
│ ├── index.ts # Deno wrapper
│ └── wasm/ # Compiled WASM module
├── rust/ # Rust source for WASM
│ ├── Cargo.toml
│ └── src/
│ └── lib.rs
├── scripts/
│ ├── build-wasm.sh # Build Rust to WASM
│ └── test-local.sh # Local testing script
├── test-client.ts # Interactive test client
└── config.toml # Supabase config
Send a message to the elizaOS agent.
Request:
{
"message": "Hello, how are you?",
"userId": "optional-user-id",
"conversationId": "optional-conversation-id"
}
Response:
{
"response": "I'm doing well, thank you for asking!",
"conversationId": "uuid-for-conversation-tracking",
"timestamp": "2025-01-10T12:00:00.000Z"
}
Headers:
| Header | Required | Description |
|---|---|---|
Authorization | Yes | Bearer YOUR_ANON_KEY or Bearer YOUR_SERVICE_ROLE_KEY |
Content-Type | Yes | application/json |
Health check endpoint.
Response:
{
"status": "healthy",
"runtime": "elizaos-deno",
"version": "2.0.0-beta.0"
}
| Variable | Required | Default | Description |
|---|---|---|---|
OPENAI_API_KEY | Yes | - | Your OpenAI API key |
OPENAI_SMALL_MODEL | No | gpt-5-mini | Small model to use |
OPENAI_LARGE_MODEL | No | gpt-5 | Large model to use |
CHARACTER_NAME | No | Eliza | Agent's name |
CHARACTER_BIO | No | A helpful AI assistant. | Agent's bio |
CHARACTER_SYSTEM | No | (default) | System prompt |
# Via CLI
supabase secrets set OPENAI_API_KEY=sk-xxx CHARACTER_NAME=MyAgent
# Via Dashboard
# Project Settings → Edge Functions → Secrets
The TypeScript implementation uses the Deno runtime natively:
// functions/eliza-chat/index.ts
import { serve } from "https://deno.land/[email protected]/http/server.ts";
import { handleChat, handleHealth } from "./lib/runtime.ts";
serve(async (req: Request): Promise<Response> => {
const url = new URL(req.url);
if (url.pathname.endsWith("/health")) {
return handleHealth();
}
if (req.method === "POST") {
return await handleChat(req);
}
return new Response("Method not allowed", { status: 405 });
});
For performance-critical operations, you can use Rust compiled to WebAssembly:
# Build WASM module
cd examples/supabase/rust
wasm-pack build --target web --out-dir ../functions/eliza-chat-wasm/wasm
# Deploy
supabase functions deploy eliza-chat-wasm
Supabase Edge Functions typically have faster cold starts than traditional Lambda:
Edge Functions run on Deno Deploy's global edge network, providing low-latency responses worldwide.
# Stream logs
supabase functions logs eliza-chat --scroll
# Get recent logs
supabase functions logs eliza-chat
View metrics and logs in:
Supabase Edge Functions pricing (as of 2025):
| Feature | Supabase Edge Functions | AWS Lambda |
|---|---|---|
| Runtime | Deno (TS/JS) | Node, Python, Rust, etc. |
| Cold Start | 50-200ms | 2-5s |
| Global Edge | ✅ Built-in | Via Lambda@Edge |
| Supabase Integration | ✅ Native | Manual |
| Python Support | ❌ | ✅ |
| Rust Support | Via WASM | Native |
Ensure the function is deployed:
supabase functions list
supabase functions deploy eliza-chat
Check your authorization header:
# Get your anon key from Supabase Dashboard
curl -H "Authorization: Bearer YOUR_ANON_KEY" ...
Set the secret:
supabase secrets set OPENAI_API_KEY=your-key
The function includes CORS headers by default. For custom domains, update the Access-Control-Allow-Origin header.
# Delete function
supabase functions delete eliza-chat
# Remove secrets
supabase secrets unset OPENAI_API_KEY