/ ディレクトリ / プレイグラウンド / Ultimate-AI-Media-Generator-Skill
← 全サーバー ↗ GitHub
● コミュニティ ZeroLu ⚡ 即起動

Ultimate-AI-Media-Generator-Skill

作者 ZeroLu · ZeroLu/Ultimate-AI-Media-Generator-Skill

Open-source image + video generation skill routing Nano Banana 2/Pro, Sora 2, Seedance, Kling, and Veo through the CyberBara API — credit preview before every call.

Lets AI agents generate images and videos without prompt-engineering expertise. Routes through CyberBara API to Nano Banana 2/Pro for images and Sora 2 / Seedance / Kling / Veo for video. Built-in prompt optimization, workflow templates (presentation / SEO content / comic), and — critically — credit visibility before submission so you can decline expensive runs. MIT licensed. Works with Claude Code, Codex, OpenClaw, Cursor, Antigravity.

▶ デモを見る 📦 インストール 💡 ユースケース

なぜ使うのか

主な機能

ライブデモ

実際の動作

ultimate-ai-media-generator-skill.replay ▶ 準備完了
0/0

インストール

クライアントを選択

~/Library/Application Support/Claude/claude_desktop_config.json  · Windows: %APPDATA%\Claude\claude_desktop_config.json
{
  "mcpServers": {
    "ultimate-ai-media-generator-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill",
        "~/.claude/skills/Ultimate-AI-Media-Generator-Skill"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop → Settings → Developer → Edit Config を開く。保存後、アプリを再起動。

~/.cursor/mcp.json · .cursor/mcp.json
{
  "mcpServers": {
    "ultimate-ai-media-generator-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill",
        "~/.claude/skills/Ultimate-AI-Media-Generator-Skill"
      ],
      "_inferred": true
    }
  }
}

Cursor は Claude Desktop と同じ mcpServers スキーマを使用。プロジェクト設定はグローバルより優先。

VS Code → Cline → MCP Servers → Edit
{
  "mcpServers": {
    "ultimate-ai-media-generator-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill",
        "~/.claude/skills/Ultimate-AI-Media-Generator-Skill"
      ],
      "_inferred": true
    }
  }
}

Cline サイドバーの MCP Servers アイコンをクリックし、"Edit Configuration" を選択。

~/.codeium/windsurf/mcp_config.json
{
  "mcpServers": {
    "ultimate-ai-media-generator-skill": {
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill",
        "~/.claude/skills/Ultimate-AI-Media-Generator-Skill"
      ],
      "_inferred": true
    }
  }
}

Claude Desktop と同じ形式。Windsurf を再起動して反映。

~/.continue/config.json
{
  "mcpServers": [
    {
      "name": "ultimate-ai-media-generator-skill",
      "command": "git",
      "args": [
        "clone",
        "https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill",
        "~/.claude/skills/Ultimate-AI-Media-Generator-Skill"
      ]
    }
  ]
}

Continue はマップではなくサーバーオブジェクトの配列を使用。

~/.config/zed/settings.json
{
  "context_servers": {
    "ultimate-ai-media-generator-skill": {
      "command": {
        "path": "git",
        "args": [
          "clone",
          "https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill",
          "~/.claude/skills/Ultimate-AI-Media-Generator-Skill"
        ]
      }
    }
  }
}

context_servers に追加。保存時に Zed がホットリロード。

claude mcp add ultimate-ai-media-generator-skill -- git clone https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill ~/.claude/skills/Ultimate-AI-Media-Generator-Skill

ワンライナー。claude mcp list で確認、claude mcp remove で削除。

ユースケース

実用的な使い方： Ultimate-AI-Media-Generator-Skill

Generate hero + inline images for a blog post

👤 Content teams needing visuals without a designer in the loop ⏱ ~15 min beginner

使うタイミング： Publishing a post and don't want stock photos.

前提条件
  • CyberBara API key — Sign up at cyberbara.ai; export CYBERBARA_API_KEY
  • Skill installed — git clone https://github.com/ZeroLu/Ultimate-AI-Media-Generator-Skill ~/.claude/skills/Ultimate-AI-Media-Generator-Skill
フロー
  1. Describe the post
    Post topic: 'retrieval-augmented generation for small teams'. Generate a hero and 3 inline images with consistent style.✓ コピーしました
    → Skill proposes model (Nano Banana Pro), shows credit preview
  2. Approve or adjust
    Approve cost — generate.✓ コピーしました
    → 4 images produced

結果： Ship-ready images with cost known upfront.

注意点
  • Inconsistent style across 4 images — Supply a style reference image; the skill supports img2img seeding
組み合わせ： wechat-article-publisher-skill · superseo-skill

Generate product shot variants for an e-commerce listing

👤 E-commerce teams ⏱ ~30 min beginner

使うタイミング： One hero shot; you need variants for A/B testing or different backgrounds.

フロー
  1. Provide source shot
    Here's product.png. Generate 5 variants: white bg, lifestyle, dark mood, in-hand, flat lay.✓ コピーしました
    → Variants proposed with credit total
  2. Iterate
    Keep variants 1 and 3; regenerate 2, 4, 5 with tighter framing.✓ コピーしました
    → Incremental iteration

結果： Listing-ready variants.

組み合わせ： bria-skill

Produce a short promo video from a storyboard

👤 Marketers needing <30s promos ⏱ ~45 min intermediate

使うタイミング： Quick promo for a social post.

フロー
  1. Write storyboard
    Storyboard: 3 shots, 5s each, 'productivity app saves team time'. Pick a model.✓ コピーしました
    → Skill picks Sora 2 or Kling based on style + credit preview
  2. Generate and review
    Approve — generate.✓ コピーしました
    → MP4 output

結果： A short video ready for social.

注意点
  • Running a 30s video at the highest model quality without checking cost — Always read the credit preview — video is the biggest spend

組み合わせ

他のMCPと組み合わせて10倍の力を

ultimate-ai-media-generator-skill + bria-skill

Generate with one provider, refine with Bria's editing primitives

Generate with Nano Banana; then background-remove with bria-skill.✓ コピーしました
ultimate-ai-media-generator-skill + wechat-article-publisher-skill

Images feed directly into WeChat drafts

After generation, publish the post to WeChat with the new hero image.✓ コピーしました

ツール

このMCPが提供する機能

ツール入力呼び出すタイミングコスト
image_generate prompt, model?, style_ref? Any still image CyberBara credits
image_edit source image, edit prompt Variants / tweaks CyberBara credits
video_generate storyboard, model? Short videos CyberBara credits (significant)
credit_preview request Always — before submit 0

コストと制限

運用コスト

APIクォータ
CyberBara per-call credits
呼び出しあたりのトークン
Low orchestration; provider cost dominates
金額
Typically cheaper than direct model APIs; still not free
ヒント
Always eyeball credit_preview — video calls can be 10–100x image calls

セキュリティ

権限、シークレット、影響範囲

認証情報の保管： CYBERBARA_API_KEY in env
データ送信先： Prompts + reference images go to CyberBara + downstream model providers

トラブルシューティング

よくあるエラーと対処法

401 invalid API key

Re-check CYBERBARA_API_KEY

確認： curl -H 'Authorization: Bearer $CYBERBARA_API_KEY' https://api.cyberbara.ai/v1/credits
Credit preview doesn't show

Update skill — credit_preview is a recent addition

Generation looks nothing like the prompt

Use the prompt-optimization helper; supply a style reference

代替案

Ultimate-AI-Media-Generator-Skill 他との比較

代替案代わりに使う場面トレードオフ
bria-skillYou need commercial-safe image editing primitives more than multi-model generationFewer generation providers

その他

リソース

📖 GitHub の公式 README を読む

🐙 オープンな issue を見る

🔍 400以上のMCPサーバーとSkillsを見る