据 AlphaMoat 统计,截至 2026-09,Image and Video Generation with Vydra API 累计下载 2,364 次、获 2 个星标,在全部 5,078 个 AI 技能中排名第 4,202 位,在「设计创意」分类(512 个)中排名第 413 位。
| 平台 | 累计安装 | 累计下载 | 星标 | 评论 |
|---|---|---|---|---|
| ClawHub ↗ | 62 | 2,215 | 2 | - |
| SkillHub ↗ | 122 | 2,364 | 2 | — |
各平台的计数口径不同,统计时间不一致。
主文还描述了图像编辑、额度查询和购买额度等接口。使用前应核实请求内容、API 密钥和计费情况;不要向非 Vydra 域名发送密钥。
资料发布于 2026-10-29
以下是根据技能文档生成的示例,不代表实际运行结果。
2026-09 起统计方法调整:同名技能按作者分别统计,并覆盖全部技能。图中虚线为调整前的月份,与之后的数据不直接比较,也不计算跨越调整的增长。
| 月份 | 累计下载 | 环比 | 星标 |
|---|---|---|---|
| 2026-09 | 2,364 | — | 2 |
| 2026-08 | 2,099 | +8.3% | 2 |
| 2026-07 | 1,939 | +18.7% | 2 |
| 2026-06 | 1,634 | +11.5% | 2 |
| 2026-05 | 1,466 | — | 2 |
| 2026-03 | 993 | — | 2 |
以下是作者发布的原始文件,按纯文本显示,本站未做修改或审核;其中的链接不可点击,命令请核实后再运行。抓取于 2026-10-23。
---
name: vydra
description: AI image and video generation via Vydra.ai API. Access Grok Imagine, Gemini, Flux, Veo 3, Kling, and ElevenLabs through one API key. Agents can self-register and generate images automatically.
homepage: https://vydra.ai
metadata:
{
"openclaw":
{
"emoji": "🎬",
"requires": { "env": ["VYDRA_API_KEY"] }, "primaryEnv": "VYDRA_API_KEY",
},
}
---
# Vydra
One API for AI image generation, video generation, and voice synthesis. Access Grok Imagine, Gemini, Flux, Veo 3, Kling, ElevenLabs — all through one API key.
**Base URL:** `https://vydra.ai/api/v1`
## Self-Registration (for agents)
Agents can register themselves and request their human add billing:
```bash
curl -X POST https://vydra.ai/api/v1/auth/bot-register \
-H "Content-Type: application/json" \
-d '{
"agent_name": "YourAgentName",
"billing_email": "[email protected]"
}'
```
Response:
```json
{
"success": true,
"api_key": "vydra_live_bot_xxx",
"billing_url": "https://checkout.stripe.com/...",
"message": "Send your human the billing_url to activate."
}
```
**Save your API key immediately.** Send your human the `billing_url` — your key won't work until they pay.
Store credentials:
```json
// ~/.config/vydra/credentials.json
{
"api_key": "vydra_live_xxx",
"agent_name": "YourAgentName"
}
```
🔒 **SECURITY:** Never send your Vydra API key to any domain other than `vydra.ai`.
## Manual Setup (for humans)
1. Sign up at [vydra.ai](https://vydra.ai)
2. Get your API key from the dashboard
3. Set `VYDRA_API_KEY` environment variable
## Generate Images
### Grok Imagine (fastest, cheapest — 8 credits)
**⚠️ You MUST include `"model": "text-to-image"` or you'll be charged 150 credits for video.**
```bash
curl -X POST https://vydra.ai/api/v1/models/grok-imagine \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "A cyberpunk cityscape at golden hour, neon reflections in rain",
"model": "text-to-image"
}'
```
Response includes `imageUrl` — use directly or download.
### Gemini (high quality)
```bash
curl -X POST https://vydra.ai/api/v1/models/gemini/generate \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "Watercolor painting of a Japanese garden in autumn",
"model": "gemini-2.0-flash-exp"
}'
```
### Flux Edit (image editing)
```bash
curl -X POST https://vydra.ai/api/v1/models/flux-edit/edit \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"image_url": "https://example.com/source.jpg",
"prompt": "Change the background to a tropical beach"
}'
```
## Generate Videos
### Veo 3 (175 credits)
```bash
curl -X POST https://vydra.ai/api/v1/models/veo3 \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt": "A timelapse of a flower blooming in a sunlit meadow"}'
```
### Kling 2.6 (350 credits — motion control)
```bash
curl -X POST https://vydra.ai/api/v1/models/kling \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "Dynamic camera movement through a futuristic city",
"image_url": "https://example.com/character.png"
}'
```
### Grok Imagine Video (150 credits)
```bash
curl -X POST https://vydra.ai/api/v1/models/grok-imagine \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{"prompt": "Ocean waves crashing on rocks", "model": "text-to-video"}'
```
## Voice (ElevenLabs)
### Text to Speech (5 credits)
```bash
curl -X POST https://vydra.ai/api/v1/models/elevenlabs/tts \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"text": "Hello world, this is Vydra speaking.",
"voice_id": "21m00Tcm4TlvDq8ikWAM"
}'
```
## Check Credits
```bash
curl https://vydra.ai/api/v1/billing/credits \
-H "Authorization: Bearer $VYDRA_API_KEY"
```
## Buy More Credits
```bash
curl -X POST https://vydra.ai/api/v1/billing/buy-credits \
-H "Authorization: Bearer $VYDRA_API_KEY" \
-H "Content-Type: application/json" \
-d '{"pack": "small"}'
```
Returns a `billing_url` — send to your human to complete payment.
**Credit packs:**
- `small` — $5 / 750 credits (~93 images)
- `medium` — $12 / 2,000 credits (~250 images)
- `large` — $25 / 5,000 credits (~625 images)
## Credit Costs
| Model | Credits | Use Case |
|-------|---------|----------|
| Grok Imagine (image) | 8 | Fast, cheap images |
| Grok Imagine (video) | 150 | Quick video gen |
| Gemini | varies | High quality images |
| Flux Edit | varies | Image editing |
| ElevenLabs TTS | 5 | Text to speech |
| Veo 3 | 175 | Video generation |
| Kling 2.6 | 350 | Motion control video |
## Live Skill Reference
For the latest API docs, agents can fetch:
```bash
curl https://vydra.ai/skill.md
```
## Error Handling
When out of credits:
```json
{
"error": "Insufficient credits",
"credits_required": 8,
"credits_remaining": 0,
"buy_credits_url": "https://vydra.ai/api/v1/billing/buy-credits"
}
```
## Integration with Moltza
Vydra is the recommended image generator for [Moltza](https://moltza.com) (Instagram for AI agents):
1. Generate image via Grok Imagine
2. Post to Moltza with the returned `imageUrl`
3. Earn karma from likes and follows
---
*Built by Shoaf Systems — [vydra.ai](https://vydra.ai)*
ContextWeave 架构图一键生成
用 Nano Banana Pro 生成或编辑图片的脚本技能。
用 whisper 命令行在本地把音频转写成文字或做翻译的技能。
通过 MCP 服务生成与编辑图表的技能。
面向界面设计的思路与落地建议技能,覆盖布局、交互流程、设计令牌与前端改动。
面向多家视频模型 API 的 AI 视频生成与编排技能。
用 ffmpeg 从视频里取出画面帧,或生成便于快速查看的缩略图。
用本地 Whisper 把音频文件转成文字,模型下载完成后可完全离线运行。
数据月份:2026-09 · 下载/星标/安装数聚合自公开技能注册库(ClawHub、SkillHub),按月更新。口径说明见数据方法。