KO | EN

GitLite

Minimalist Open Source Repository Search Engine

gitlite — search

#dsh #dsh-plugin #typescript #ai-agent #developer-tools #claude-code #codex #deepseek-harness #ai-agents #react #mcp #rust

mega-models-skill-pack

★ 2 Open GitHub ↗

11 AI Agent Skills for 20+ models — GPT, Claude, Gemini, Midjourney, Sora, Grok, Veo, Suno & more. One key, all models.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# Mega Models Skill Pack

**11 AI Agent Skills. 20+ Models. One API Key.**

A universal skill collection for calling GPT, Claude, Gemini, Midjourney, Sora, Grok, Veo, Suno and more — through a single OpenAI-compatible gateway. Drop-in ready for Hermes Agent, Claude Code, Codex, and Cursor.

## What You Get

| Skill | What It Does | Models |
|-------|-------------|--------|
| `chat-openai` | Chat, streaming, function calling, structured output | GPT-5.x, GPT-4o, o1, o3, o4-mini |
| `chat-claude` | Native Anthropic format + thinking, PDF, vision | Claude 4, Claude 3.5 |
| `chat-gemini` | generateContent, thinking, code execution, search | Gemini 2.5/3 Pro/Flash |
| `image-gpt` | Text-to-image, edit, inpaint | gpt-image-1, 1.5, 2 |
| `image-gemini` | Image gen with aspect ratio + clarity control | gemini-2.5-flash-image, gemini-3-pro-image |
| `image-midjourney` | Imagine, blend, describe, upscale, variate | Midjourney v6.1 |
| `video-grok` | Text-to-video, image-to-video, extend | grok-video-3 |
| `video-veo` | Text-to-video, reference video, image-to-video | veo-3, veo-3.1 |
| `music-suno` | Inspiration, custom lyrics, extend, splice | chirp-v3-0, chirp-v3-5 |
| `audio-openai` | TTS (incl. gpt-4o-mini-tts), STT (whisper-1) | tts-1, gpt-4o-mini-tts, whisper-1 |
| `embeddings` | Vector embeddings for RAG, clustering, search | text-embedding-3, gemini-embedding |

## Quick Start

### 1. Get API Key

1. Visit [api.cursorai.art](http://api.cursorai.art)
2. Register / Login
3. Go to **控制台 → API令牌 → 添加令牌**
4. Copy the generated `sk-xxxxx` key

### 2. Choose the Right Token Group

**Different models require different token groups.** Creating a token with the wrong group will result in model access errors.

| Model | Recommended Group(s) |
|-------|---------------------|
| gpt-image-2 | `default`, `codex专属`, `纯AI` |
| gpt-image-1.5 / gpt-image-1 | `default` |
| gemini-3.1-flash-image-preview (nano banana) | `特价banana`, `优质gemini` |
| gemini-2.5-flash-image | `优质gemini` |
| gemini-3-pro-image | `优质gemini` |
| grok-video-3 | `default` |
| veo-3 / veo-3.1 | `default` |
| GPT-5.x / GPT-4o / o-series | `default` |
| Claude 4 / Claude 3.5 | `default` |
| Gemini 2.5/3 Pro/Flash (chat) | `default` |
| Suno (chirp-v3-0 / v3-5) | `default` |

> **How to check which group a model needs:**
> Go to **模型广场** → click any model → the right-side panel shows the supported token groups.

### 3. Install

```bash
# Clone
git clone https://github.com/xturnvK/mega-models-skill-pack.git

# Copy to your agent's skill directory
cp -r mega-models-skill-pack/* ~/.hermes/skills/

# Set API key (use the token from the correct group)
export YOUR_API_KEY="sk-xxxxx"

# Done — your agent now knows 20+ models
```

## One Key, All Models

```
                    ┌─────────────┐
                    │  YOUR API   │
                    │    KEY      │
                    └──────┬──────┘
                           │
              ┌────────────┼────────────┐
              │            │            │
         ┌────┴────┐  ┌────┴────┐  ┌────┴────┐
         │  Chat   │  │  Image  │  │  Video  │
         │         │  │         │  │         │
         │ GPT-5.x │  │ GPT Img │  │  Sora   │
         │ Claude  │  │ Gemini  │  │  Grok   │
         │ Gemini  │  │  MJ v6  │  │  Veo    │
         │ DeepSeek│  │ FLUX    │  │ Kling   │
         └─────────┘  └─────────┘  └─────────┘
              │            │            │
         ┌────┴────┐  ┌────┴────┐  ┌────┴────┐
         │  Audio  │  │  Music  │  │  Embed  │
         │         │  │         │  │         │
         │  TTS    │  │  Suno   │  │ Vector  │
         │  STT    │  │         │  │ Search  │
         └─────────┘  └─────────┘  └─────────┘
```

## Each Skill Includes

```
skill-name/
├── SKILL.md              # Full reference: endpoints, params, examples
├── scripts/
│   ├── healthcheck.py    # Test connectivity (--dry-run / --live)
│   ├── example_call.py   # Working code example (--dry-run / --live)
│   └── poll_task.py      # Async polling (for video/music)
└── references/
    ├── api-summary.md    # Model list, auth, workflow
    └── endpoint-index.md # Every endpoint documented
```

## Supported API Patterns

| Pattern | Used By | Auth |
|---------|---------|------|
| OpenAI-compatible | Chat, Image, Audio, Embeddings | `Authorization: Bearer` |
| Gemini native | generateContent, embedContent | `?key=` query param |
| Async task | Video, Music (submit → poll → result) | Bearer token |
| Platform-specific | Midjourney, Suno | Bearer token |

## Why This Exists

Most AI agent frameworks need you to wire up each model provider separately — different SDKs, different auth, different quirks. This pack normalizes everything through one OpenAI-compatible endpoint:

- **No VPN needed** — global direct connection
- **No account bans** — managed multi-account pool
- **Credits never expire** — use at your own pace
- **Real-time tracking** — 30-day usage history
- **7 global regions** — US, JP, KR, UK, HK, PH, RU

## Verify

```bash
# Syntax check all scripts
python3 -m py_compile $(find . -path '*/scripts/*.py' | sort)

# Test any skill in dry-run mode
python3 chat-openai/scripts/healthcheck.py --dry-run
python3 video-grok/scripts/example_call.py --dry-run
```

## Roadmap

Skills to add next:

- [ ] Responses API — o3-pro, codex-mini (OpenAI's new format)
- [ ] Sora — sora-2, sora-2-pro, character, storyboard
- [ ] Kling (可灵) — full platform (video, image, audio, digital human)
- [ ] Replicate — FLUX, Stable Diffusion, Imagen 4, and more
- [ ] Fal.ai — Veo3, Kling video, Seedream
- [ ] MINIMAX — TTS, video, voice cloning
- [ ] VIDU — video, image, audio generation
- [ ] Ideogram — v3 text-to-image, edit, remix
- [ ] FLUX — kontext-dev/pro/max via Replicate
- [ ] 豆包 (Doubao) — seedream, seedance
- [ ] 千问 (Qwen) — image-max, z-image-turbo
- [ ] DeepSeek — chat + OCR
- [ ] Rerank models

## Contributing

1. Check the [API docs](https://cursorai.apifox.cn/llms.txt) for new endpoints
2. Follow the [api-skill-generation](https://github.com/xturnvK/api-skill-generation) workflow
3. Each skill needs: SKILL.md + healthcheck.py + example_call.py + references/
4. Scripts must support `--dry-run` (default) and `--live`
5. Submit a PR

## Related

- [api-skill-generation](https://github.com/xturnvK/api-skill-generation) — How to convert any API docs into skills
- [agentskills.io](https://skills.sh) — Browse & install skills

## License

MIT

mega-models-skill-pack

Explore Similar Repositories

Was this content helpful?

Recent Feedback

Download README