Search

Note: When changing a SQLite table's schema, change the table's name (e.g., add _2 or _3) to create a fresh table.
### OpenAI
```ts
import { OpenAI } from "https://esm.town/v/std/openai";
const openai = new OpenAI();
const completion = await openai.chat.completions.create({
  messages: [
    { role: "user", content: "Say hello in a creative way" },

Note: When changing a SQLite table's schema, change the table's name (e.g., add _2 or _3) to create a fresh table.
### OpenAI
```ts
import { OpenAI } from "https://esm.town/v/std/openai";
const openai = new OpenAI();
const completion = await openai.chat.completions.create({
  messages: [
    { role: "user", content: "Say hello in a creative way" },

const openrouterApiKey2 = Deno.env.get("OPENROUTER_API_KEY2");
// Poe API (new OpenAI-compatible format)
const poeApiKey = Deno.env.get("POE_API_KEY");
];
// Poe vision models (new API - OpenAI compatible)
const POE_VISION_MODELS = [
  "Gemini-2.5-Flash-Lite",
];
// ============ POE API (NEW OpenAI-compatible format) ============
async function callPoeApi(
/**
 * Vision pipeline:
 *  1) Poe Vision models (primary - new OpenAI-compatible API)
 *  2) NVIDIA Vision models (fallback)
 *  3) OpenRouter Vision models (final fallback)
  prompt: string
): Promise<string> {
  // 1. Try Poe Vision first (new OpenAI-compatible API)
  console.log("Trying Poe vision models first...");
  const poeResult = await analyzeImageWithPoe(imageBase64, prompt);

import { OpenAI } from "https://esm.sh/openai@4.20.1";
export default async function handler(req: Request): Promise<Response> {
    console.log(`📸 Processing ${base64Images.length} image(s)`);
    const apiKey = Deno.env.get("OPENAI_API_KEY");
    if (!apiKey) {
      console.error("❌ OPENAI_API_KEY not set");
      return new Response(
        JSON.stringify({
    }
    const openai = new OpenAI({ apiKey });
    console.log("🤖 Calling OpenAI for fish identification...");
    // Language-specific instruction
    }
    const response = await openai.chat.completions.create({
      model: "gpt-4o",
      messages: [
    if (!content) {
      throw new Error("No response from OpenAI");
    }
    if (error instanceof Error) {
      if (
        error.message === "No response from OpenAI" ||
        error.message === "Invalid response format from AI"
      ) {

| `TODOS_DB_ID`    | Database ID where todos are synced (32-char ID without hyphens)                       |
### Recommended: OpenAI API Key
| Variable         | Description                                           |
| ---------------- | ----------------------------------------------------- |
| `OPENAI_API_KEY` | OpenAI API key for AI features (strongly recommended) |
**What AI does:**
- Disambiguates when multiple contacts match a name
**Without `OPENAI_API_KEY`:** Falls back to Val Town's shared OpenAI client (10 req/min limit). If AI is unavailable, the system still works but uses raw text, heading/@mention-only owner resolution, and default due dates.
### Required Database Properties
Visit the root URL of your val (`main.http.tsx`) to see a dashboard showing:
- **Connection status**: Notion API and OpenAI connectivity
- **Property mappings**: Which properties are configured and whether they exist in your database
| ----------------------- | --------------------------------------------- |
| `NOTION_WEBHOOK_SECRET` | API key for protecting endpoints              |
| `OPENAI_API_KEY`        | For AI features (see Quick Start for details) |
---

  options = {},
) => {
  // Initialize OpenAI API stub with custom configuration for iFlow
  const { Configuration, OpenAIApi } = await import(
    "https://esm.sh/openai@3.3.0"
  );
  const configuration = new Configuration({
    apiKey: process.env.OPENAI_API_KEY || process.env.OPENAI,
    basePath: process.env.OPENAI_BASE_URL || undefined,
  });
  const openai = new OpenAIApi(configuration);
  // Request chat completion
  const messages = typeof prompt === "string"
    : prompt;
  const model = process.env.MODEL_NAME || "gpt-3.5-turbo-0613";
  const { data } = await openai.createChatCompletion({
    model,
    messages: messages as any,

  handler
- Environment variables via `Deno.env.get()`
- Val Town std libraries used: `email` (forwarding), `openai` (LLM fallback)
## Code Standards
- **Geocoder**: US Census Bureau (`geocoding.geo.census.gov`) — free, no API
  key, handles DC intersections
- **LLM fallback**: OpenAI gpt-4o-mini via Val Town `std/openai` for locations
  that resist deterministic parsing
- **Email format**: Location is always in

  intersections well)
- **LLM fallback**: When deterministic parsing can't produce a geocodable
  address, OpenAI gpt-4o-mini rewrites the location string before a second
  geocode attempt
- **Email parsing**: Location is extracted from the pattern

_2 or _3) to create a fresh table.
### OpenAI
```ts
import { OpenAI } from "https://esm.town/v/std/openai";
const openai = new OpenAI();
const completion = await openai.chat.completions.create({
  messages: [
    { role: "user", content: "Say hello in a creative way" },

async function geocodeWithLLM(rawLocation: string): Promise<GeoResult | null> {
  const { OpenAI } = await import("https://esm.town/v/std/openai");
  const openai = new OpenAI();
  const completion = await openai.chat.completions.create({
    messages: [
      {