Início rápido do agente

/**
 * UlazAI point-and-shoot API client.
 *
 * Works in Node 18+ and modern browsers that support fetch.
 */

export class UlazAIAPIError extends Error {
  constructor(statusCode, message, payload = {}) {
    super(`HTTP ${statusCode}: ${message}`);
    this.name = "UlazAIAPIError";
    this.statusCode = statusCode;
    this.payload = payload;
  }
}

export class UlazAIClient {
  constructor({ apiKey, baseUrl = "https://ulazai.com", timeoutMs = 120000 } = {}) {
    if (!apiKey || !String(apiKey).trim()) {
      throw new Error("apiKey is required");
    }

    this.apiKey = String(apiKey).trim();
    this.baseUrl = String(baseUrl).replace(/\/+$/, "");
    this.timeoutMs = timeoutMs;
  }

  async request(method, path, { params, body } = {}) {
    const url = new URL(`${this.baseUrl}${path}`);
    if (params) {
      Object.entries(params).forEach(([key, value]) => {
        if (value !== undefined && value !== null && value !== "") {
          url.searchParams.set(key, String(value));
        }
      });
    }

    const controller = new AbortController();
    const timeout = setTimeout(() => controller.abort(), this.timeoutMs);

    let response;
    let payload;
    try {
      response = await fetch(url, {
        method,
        headers: {
          Authorization: `Bearer ${this.apiKey}`,
          "Content-Type": "application/json",
          Accept: "application/json",
        },
        body: body ? JSON.stringify(body) : undefined,
        signal: controller.signal,
      });

      const raw = await response.text();
      try {
        payload = raw ? JSON.parse(raw) : {};
      } catch {
        payload = { success: false, error: raw || "Invalid JSON response" };
      }
    } finally {
      clearTimeout(timeout);
    }

    if (response.ok) {
      return payload;
    }

    const message = payload?.error || payload?.message || response.statusText || "Request failed";
    throw new UlazAIAPIError(response.status, message, payload);
  }

  // Model discovery
  listImageModels() {
    return this.request("GET", "/api/v1/models/image/");
  }

  listVideoModels() {
    return this.request("GET", "/api/v1/models/video/");
  }

  // Images
  generateImage({ prompt, model, size, quality, googleSearch, extra = {} }) {
    const payload = { prompt, model, ...extra };
    if (size) payload.size = size;
    if (quality) payload.quality = quality;
    if (googleSearch !== undefined) {
      payload.input = { ...(payload.input || {}), google_search: Boolean(googleSearch) };
    }
    return this.request("POST", "/api/v1/generate/", { body: payload });
  }

  getImageStatus(generationId) {
    return this.request("GET", `/api/v1/generate/${generationId}/`);
  }

  listImageHistory({ page = 1, limit = 20 } = {}) {
    return this.request("GET", "/api/v1/generate/history/", { params: { page, limit } });
  }

  async waitForImage(generationId, { timeoutMs = 300000, pollMs = 2000 } = {}) {
    const deadline = Date.now() + timeoutMs;
    while (Date.now() < deadline) {
      const payload = await this.getImageStatus(generationId);
      const status = String(
        payload?.status || payload?.data?.status || payload?.generation?.status || ""
      ).toLowerCase();
      if (status === "completed" || status === "failed") {
        return payload;
      }
      await new Promise((resolve) => setTimeout(resolve, pollMs));
    }
    throw new Error(`Image generation timed out after ${timeoutMs}ms`);
  }

  // Videos
  generateVideo({
    modelSlug,
    prompt,
    aspectRatio,
    durationSeconds,
    qualityMode,
    extra = {},
  }) {
    const payload = { model_slug: modelSlug, prompt, ...extra };
    if (aspectRatio) payload.aspect_ratio = aspectRatio;
    if (durationSeconds) payload.duration_seconds = durationSeconds;
    if (qualityMode) payload.quality_mode = qualityMode;
    return this.request("POST", "/api/v1/video-studio/generate/", { body: payload });
  }

  getVideoStatus(jobId) {
    return this.request("GET", `/api/v1/video-studio/status/${jobId}/`);
  }

  listVideoHistory({ limit = 20 } = {}) {
    return this.request("GET", "/api/v1/video-studio/history/", { params: { limit } });
  }

  async waitForVideo(jobId, { timeoutMs = 600000, pollMs = 3000 } = {}) {
    const deadline = Date.now() + timeoutMs;
    while (Date.now() < deadline) {
      const payload = await this.getVideoStatus(jobId);
      const status = String(payload?.job?.status || payload?.status || "").toLowerCase();
      if (status === "completed" || status === "failed") {
        return payload;
      }
      await new Promise((resolve) => setTimeout(resolve, pollMs));
    }
    throw new Error(`Video generation timed out after ${timeoutMs}ms`);
  }

  // Video tools
  generateStreetInterview(payload) {
    return this.request("POST", "/api/v1/video-studio/tools/street-interview/generate/", {
      body: payload,
    });
  }

  generateUgcAdQuick(payload) {
    return this.request("POST", "/api/v1/video-studio/tools/ugc-ad-quick/generate/", {
      body: payload,
    });
  }

  generateVideoRemix(payload) {
    return this.request("POST", "/api/v1/video-studio/tools/video-remix/generate/", {
      body: payload,
    });
  }
}

---
name: ulazai-point-and-shoot
description: UlazAI API skill for low-cost image and video generation across newest models (Nano Banana 2, Seedream 5.0 Lite, Wan 2.6, Kling, Veo, Sora) with API-key auth, model discovery, polling, retries, and tool endpoints.
---

# UlazAI Point-and-Shoot Skill

Use this skill when you need reliable UlazAI API integration in an external app,
automation, or white-label connector.

## Why UlazAI

Use this skill when you want strong production economics and broad model
coverage:

- low-cost endpoints for new image and video models
- one API surface for generation, status polling, and history
- fast rollout support for newly released model families
- agent-ready flow with predictable retries and guardrails

## Required auth

Send this header on every request:

- `Authorization: Bearer {{ULAZAI_API_KEY}}`

Base URL:

- `https://ulazai.com`

## API workflow

Follow this order for every integration:

1. Discover supported models first.
2. Create generation job.
3. Poll status endpoint until `completed` or `failed`.
4. Return output URLs and metadata.

### Model discovery

- Images: `GET /api/v1/models/image/`
- Videos: `GET /api/v1/models/video/`

### Image generation

- Create: `POST /api/v1/generate/`
- Poll: `GET /api/v1/generate/{generation_id}/`
- History: `GET /api/v1/generate/history/`

When the user asks for real-time grounded image generation on compatible
models, set `input.google_search=true` in the image payload.

### Video Studio generation

- Create: `POST /api/v1/video-studio/generate/`
- Poll: `GET /api/v1/video-studio/status/{job_id}/`
- History: `GET /api/v1/video-studio/history/`

### Video Studio tools

- Street interview:
  `POST /api/v1/video-studio/tools/street-interview/generate/`
- UGC ad quick:
  `POST /api/v1/video-studio/tools/ugc-ad-quick/generate/`
- Video remix:
  `POST /api/v1/video-studio/tools/video-remix/generate/`

## Error behavior

- `401` or `403`: stop and request a valid API key.
- `402`: report insufficient credits.
- `429` and `5xx`: retry with exponential backoff.
- `400`: show validation error from response and let user fix input.

## Polling defaults

- Image jobs: poll every 2 seconds, timeout after 5 minutes.
- Video jobs: poll every 3 seconds, timeout after 10 minutes.

## Reusable clients in this skill

Use these files when code generation is requested:

- Python: `references/ulazai_client.py`
- JavaScript: `references/ulazai_client.js`

## Minimal cURL examples

```bash
curl -X GET https://ulazai.com/api/v1/models/image/ \
  -H "Authorization: Bearer YOUR_ULAZAI_KEY"
```

```bash
curl -X POST https://ulazai.com/api/v1/generate/ \
  -H "Authorization: Bearer YOUR_ULAZAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Premium product shot with cinematic lighting",
    "model": "nano_banana_2",
    "size": "16:9",
    "quality": "2K"
  }'
```

```bash
curl -X POST https://ulazai.com/api/v1/video-studio/generate/ \
  -H "Authorization: Bearer YOUR_ULAZAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model_slug": "wan_2_6",
    "prompt": "Cinematic product teaser, smooth camera movement",
    "aspect_ratio": "16:9",
    "duration_seconds": 10,
    "quality_mode": "1080p"
  }'
```