Ultimate AI Media Generator Skill

Overview

A powerful skill for generating AI images and videos using the world's leading generative models:

• - Sora 2 — OpenAI's flagship video generation model
• Kling 2.6 — High-quality video generation with image-to-video support
• Seedance 2.0 — ByteDance's advanced video model
• Nano Banana Pro — Fast, cost-effective image generation
• Veo 3.1 — Google's state-of-the-art video model
• And more...

Supports all major generation workflows: text-to-image, text-to-video, image-to-image, image-to-video, and video-to-video. The skill handles credit estimation, task creation, status polling, and automatic media output saving.

Built-in Prompt Optimization

This skill automatically optimizes prompts for specific use cases to achieve the best results:

• - SEO Articles — Generate featured images and illustrations optimized for blog posts and articles
• Presentations/Slides — Create professional visuals for PowerPoint, Keynote, and Google Slides
• Anime/Manga — Generate anime-style artwork with optimized prompts for consistent style
• Product Photography — Create product shots with proper lighting and composition
• Social Media — Generate platform-optimized visuals for Instagram, TikTok, YouTube
• Marketing Materials — Create ad creatives, banners, and promotional content

The included workflow templates (workflows/ folder) provide ready-to-use prompts and best practices for each use case.

Implementation Architecture

The runtime uses a layered Python architecture:

• - scripts/cyberbara_api.py: thin entrypoint only
• INLINECODE2: command parsing and command routing
• INLINECODE3: flow orchestration (generation + polling)
• INLINECODE4: safety and policy rules (credits quote + formal confirmation)
• INLINECODE5: raw CyberBara API client
• INLINECODE6: API key discovery and local persistence
• INLINECODE7: fixed base URL and shared constants

When extending behavior, keep business rules in usecases/ or policies/, not in scripts/.

Set Up Runtime Inputs

The script uses fixed base URL:

CODEBLOCK0

API key lookup order:

1. 1. INLINECODE11
2. environment variable INLINECODE12
3. local cache file INLINECODE13
4. interactive prompt (if running in terminal)

Recommended one-time setup command:

CODEBLOCK1

Or save from environment variable:

CODEBLOCK2

If API key is missing, the script immediately asks for it and shows where to get one:

CODEBLOCK3

When you provide API key via --api-key or interactive prompt, it is saved to:

CODEBLOCK4

Future runs reuse this cached key, so users do not need to provide it every time.

Run The Standard Generation Flow

1. 1. Discover available models.
2. Upload reference images when task scene needs image inputs.
3. Quote credits before creating a generation task.
4. Create image or video generation task and wait for final output.
5. Automatically save generated media locally and open it.
6. Check usage records when needed.

Reference commands:

CODEBLOCK5

Image and video generation are confirmation-gated by default:

CODEBLOCK6

INLINECODE15 format:

CODEBLOCK7

Only use --yes after explicit user approval has been obtained:

CODEBLOCK8

Control auto-save and open behavior:

CODEBLOCK9

Use Script Capabilities

INLINECODE17 supports:

• - setup-api-key to persist API key into local cache
• INLINECODE19 to list public models (--media-type image|video optional)
• INLINECODE21 to upload local image files to INLINECODE22
• INLINECODE23 to estimate credit cost from JSON request body
• INLINECODE24 to auto-quote credits, compute total for batch requests, require formal confirmation, create task(s), wait, then save/open outputs
• INLINECODE25 to auto-quote credits, compute total for batch requests, require formal confirmation, create task(s), wait, then save/open outputs
• INLINECODE26 to fetch task snapshot by task ID
• INLINECODE27 to poll task until success, failed, or canceled, then save/open outputs
• INLINECODE31 and usage to inspect credits
• INLINECODE33 for direct custom endpoint calls

Use --file request.json instead of --json for long payloads.

Enforce API Payload Rules

• - Send auth via API key (Authorization: Bearer <key> or x-api-key).
• Send public request fields under options.* only.
• Prefer explicit scene to avoid inference ambiguity.
• Include options.image_input for image-to-image and image-to-video.
• Include options.video_input for video-to-video.
• Poll /api/v1/tasks/{taskId} until final status; only success guarantees output URLs.
• Before every image or video generation submission, obtain quote first and get explicit user confirmation.
• For multiple image/video requests, calculate and present total estimated credits before submission.
• Save output files under media_outputs/ by default and auto-open them unless disabled.

Navigate Detailed Model Options

Use the reference file for full model matrices and examples:

• - INLINECODE48

For fast lookup in large reference:

CODEBLOCK10

Handle Common Failures

• - invalid_api_key or api_key_required: verify key and headers.
• INLINECODE51: quote first or recharge credits.
• INLINECODE52 or scene_not_supported: choose scene supported by model.
• INLINECODE54: verify prompt and options requirements by model.
• INLINECODE57: verify task ID and environment domain.