Search

Code
3,897

1 matches

const urls = [
  "https://console.groq.com/docs/quickstart",
  "https://console.groq.com/docs/openai",
  "https://console.groq.com/docs/models",
  "https://console.groq.com/docs/rate-limits",

yawnxyz/groq-docs/AGENTS.md

4 matches

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" },

flymaster/lifehub/AGENTS.md

4 matches

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" },

realtime/Bee/main.ts

8 matches

// @ts-ignore
import { OpenAI } from "https://esm.town/v/std/openai?v=4";
// @ts-ignore
import { blob } from "https://esm.town/v/std/blob?v=11";
 *
 * Notes:
 * - Uses std/openai (Val credentials) and std/blob for persistence
 * - UI polls and triggers small "work ticks" to progress jobs
 * - Keep prompts deterministic, temperature low, JSON-only outputs
   ========================= */
const OPENAI_MODEL = "gpt-4o-mini"; // adjust as preferred
const JSON_RESPONSE_FORMAT = { type: "json_object" as const };
const EXTRACTION_CONCURRENCY_PER_TICK = 4; // 24 lenses processed over ~6 ticks
   ========================= */
const client = new OpenAI();
/** Persist and load helpers */
  const resp = await client.chat.completions.create({
    model: OPENAI_MODEL,
    temperature: 0.1,
    response_format: JSON_RESPONSE_FORMAT,
  const resp = await client.chat.completions.create({
    model: OPENAI_MODEL,
    temperature: 0.2,
    response_format: JSON_RESPONSE_FORMAT,
  const resp = await client.chat.completions.create({
    model: OPENAI_MODEL,
    temperature: 0.1,
    response_format: JSON_RESPONSE_FORMAT,
  const resp = await client.chat.completions.create({
    model: OPENAI_MODEL,
    temperature: 0.2,
    response_format: JSON_RESPONSE_FORMAT,

damixnr/OpenCloset/README.md

4 matches


- **Backend**: Hono + MCP Lite + Drizzle ORM + SQLite + Cloudinary
- **Widget**: React 19 + TanStack Router + OpenAI App SDK
## Quick Start
## Message Scoping
Messages are automatically scoped using the `openai/subject` field that ChatGPT includes in request metadata. This provides authless data isolation - each subject gets its own message board.
The scoping happens in tool handlers:
```typescript
const subject = ctx.request.params._meta?.["openai/subject"];
const messages = await getMessages(subject);
```
The exact semantics of `openai/subject` are determined by ChatGPT.
## Where do I go from here?

damixnr/OpenCloset/server.ts

50 matches

function widgetMeta(invoking?: string, invoked?: string) {
  return {
    "openai/outputTemplate": WIDGET_URI,
    "openai/toolInvocation/invoking": invoking,
    "openai/toolInvocation/invoked": invoked,
    "openai/widgetAccessible": true,
    "openai/resultCanProduceWidget": true,
  } as const;
}
          _meta: {
            ...widgetMeta(),
            "openai/widgetCSP": {
              connect_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
              resource_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
          _meta: {
            ...widgetMeta(),
            "openai/widgetCSP": {
              connect_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
              resource_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
          _meta: {
            ...widgetMeta(),
            "openai/widgetCSP": {
              connect_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
              resource_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
          _meta: {
            ...widgetMeta(),
            "openai/widgetCSP": {
              connect_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
              resource_domains: [baseUrl, "https://esm.sh", "https://cdn.jsdelivr.net"],
  _meta: widgetMeta("Loading messages...", "Messages loaded"),
  handler: async (args, ctx) => {
    const subject = ctx.request.params._meta?.["openai/subject"];
    if (!subject) {
      throw new Error("Missing openai/subject in request metadata");
    }
  _meta: widgetMeta("Loading message...", "Message loaded"),
  handler: async ({ id }, ctx) => {
    const subject = ctx.request.params._meta?.["openai/subject"];
    if (!subject) {
      throw new Error("Missing openai/subject in request metadata");
    }
    }
    const subject = ctx.request.params._meta?.["openai/subject"];
    if (!subject) {
      throw new Error("Missing openai/subject in request metadata");
    }
  _meta: {
    ...widgetMeta("Loading item...", "Item loaded"),
    "openai/outputTemplate": ITEM_CARD_URI,
  },
  handler: async ({ category }) => {
      _meta: {
        ...widgetMeta(),
        "openai/outputTemplate": ITEM_CARD_URI,
      },
    };
  _meta: {
    ...widgetMeta("Loading items...", "Items loaded"),
    "openai/outputTemplate": MULTI_ITEM_CARD_URI,
  },
  handler: async ({ count }) => {
      _meta: {
        ...widgetMeta(),
        "openai/outputTemplate": MULTI_ITEM_CARD_URI,
      },
    };
  _meta: {
    ...widgetMeta("Creating outfit...", "Outfit ready"),
    "openai/outputTemplate": OUTFIT_CARD_URI,
  },
  handler: async () => {
      _meta: {
        ...widgetMeta(),
        "openai/outputTemplate": OUTFIT_CARD_URI,
      },
    };
  outputSchema: CaptureItemsOutput,
  _meta: {
    "openai/toolInvocation/invoking": "Scanning & uploading…",
    "openai/toolInvocation/invoked": "Saved to your closet.",
    "openai/widgetAccessible": true,
    "openai/resultCanProduceWidget": true,
  },
  handler: async (args) => {
      structuredContent: isMultiple ? { items: added } : { item: added[0] },
      _meta: {
        "openai/outputTemplate": isMultiple ? MULTI_ITEM_CARD_URI : ITEM_CARD_URI,
        "openai/toolInvocation/invoking": "Scanning & uploading…",
        "openai/toolInvocation/invoked": "Saved to your closet.",
        "openai/widgetAccessible": true,
        "openai/resultCanProduceWidget": true,
      },
    };
  outputSchema: SuggestOutfitOutput,
  _meta: {
    "openai/outputTemplate": OUTFIT_CARD_URI,
    "openai/toolInvocation/invoking": "Picking pieces…",
    "openai/toolInvocation/invoked": "Outfit ready.",
    "openai/widgetAccessible": true,
    "openai/resultCanProduceWidget": true,
  },
  handler: async () => {
      structuredContent: { top, bottom },
      _meta: {
        "openai/outputTemplate": OUTFIT_CARD_URI,
        "openai/toolInvocation/invoking": "Picking pieces…",
        "openai/toolInvocation/invoked": "Outfit ready.",
        "openai/widgetAccessible": true,
        "openai/resultCanProduceWidget": true,
      },
    };
  }),
  _meta: {
    "openai/outputTemplate": MULTI_ITEM_CARD_URI,
    "openai/toolInvocation/invoking": "Loading closet…",
    "openai/toolInvocation/invoked": "Closet loaded.",
    "openai/widgetAccessible": true,
    "openai/resultCanProduceWidget": true,
  },
  handler: async ({ category }) => {
      structuredContent: { items },
      _meta: {
        "openai/outputTemplate": MULTI_ITEM_CARD_URI,
        "openai/toolInvocation/invoking": "Loading closet…",
        "openai/toolInvocation/invoked": "Closet loaded.",
        "openai/widgetAccessible": true,
        "openai/resultCanProduceWidget": true,
      },
    };

krzksz/basic-html-starter/AGENTS.md

4 matches

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" },

damixnr/OpenCloset/OPENCLOSET.md

2 matches


### Widget Integration
- Each tool specifies `openai/outputTemplate` in metadata
- Widgets automatically render based on structured content
- Consistent styling across all clothing-related widgets
1. **Cloudinary upload fails**: Check environment variables
2. **No outfit suggestions**: Ensure both tops and bottoms are in closet
3. **Widget not displaying**: Verify `openai/outputTemplate` metadata
### Debug Tips

damixnr/OpenCloset/TESTING.md

1 matches

- All widgets use the `useToolOutput` hook to access structured data
- Theme support via `useTheme` hook (light/dark mode)
- Interactive actions send follow-up messages to ChatGPT via `getOpenAI().sendFollowUpMessage()`
### Styling

damixnr/OpenCloset/WIDGET_REGISTRATION.md

3 matches

- Displays a single wardrobe item with image and category
- Uses inline styles for consistent appearance
- Accesses data via `globalThis.openai?.toolOutput?.structuredContent`
### 2. MultiItemCard (`/frontend/widgets/components/MultiItemCard.tsx`)
### MCP Integration
- Proper `_meta` configuration with `openai/outputTemplate`
- Structured content matching component expectations
- Error handling for missing data
### Data Access Pattern
- Uses `globalThis.openai?.toolOutput?.structuredContent`
- Direct access to structured data (no hooks needed)
- Matches the original component specifications exactly