name: visioncraft description: | Generate images and videos via Agnes AI. Triggers on any request to create, draw, render, design, edit, restyle, or animate visual content — pictures, posters, illustrations, product visuals, photos, or video clips. Handles text-to-image, image-to-image, text-to-video, image-to-video, multi-image video, and keyframe animation.
VisionCraft
Generate images and videos via Agnes AI APIs. Calls PowerShell scripts in scripts/. API key is read from environment variable AGNES_API_KEY.
First time? See Prerequisites. For prompt-writing tips and parameter details, see
references/.
Quick Reference
| Task | Command |
|---|---|
| Text-to-Image | scripts/gen_image.ps1 -Prompt "..." |
| Image-to-Image | scripts/gen_image.ps1 -Prompt "..." -ImageUrl "https://..." |
| Text-to-Video | scripts/gen_video.ps1 -Prompt "..." |
| Image-to-Video | scripts/gen_video.ps1 -Prompt "..." -Image "https://..." |
| Multi-image Video | scripts/gen_video.ps1 -Prompt "..." -Image "https://a.png","https://b.png" |
| Keyframes Animation | scripts/gen_video.ps1 -Prompt "..." -Image "https://a.png","https://b.png" -Mode keyframes |
Defaults: Images 2048x2048. Videos 1920x1080 @ 60fps, 241 frames (~4s).
Prerequisites
Set the API key once (persists across sessions):
[System.Environment]::SetEnvironmentVariable("AGNES_API_KEY", "sk-your-key-here", "User")
Or for the current session only:
$env:AGNES_API_KEY = "sk-your-key-here"
Verify with echo $env:AGNES_API_KEY.
How to Use This Skill
1. Identify the request
| User says | Mode |
|---|---|
| "make/draw/generate a picture of..." | Text-to-Image |
| "edit/transform/restyle this image..." (URL provided) | Image-to-Image |
| "make a video / animate / generate a clip of..." | Text-to-Video |
| "animate this photo / make this image move" (1 image) | Image-to-Video |
| "blend these images into a video" (multiple images) | Multi-image Video |
| "smooth transition / morph from A to B" (2+ images) | Keyframes Animation |
2. Refine the prompt if needed
If the request is vague, briefly suggest a richer prompt before running. For prompt structure and best practices see references/prompt-guide.md.
3. Run the script
Pass aspect ratio with -Ratio and duration with -Seconds for the simplest control. See references/parameters.md for all parameters.
# Examples:
scripts/gen_image.ps1 -Prompt "..." -Ratio 16:9
scripts/gen_video.ps1 -Prompt "..." -Ratio 9:16 -Seconds 8
4. Surface the result
The scripts print a MEDIA:<url> line on success. Always echo the URL on its own line — never bury it in a paragraph. If the platform supports inline media rendering, embed it; otherwise the URL alone is enough.
Never download the result to local disk unless the user explicitly asks.
Output Format
Image:
✅ Image generated.
MEDIA:https://platform-outputs.agnes-ai.space/images/...
Video:
🎬 Video generated (4.0s, 1920x1080).
MEDIA:https://platform-outputs.agnes-ai.space/videos/...
Errors & Troubleshooting
| Error | Fix |
|---|---|
AGNES_API_KEY not set | Set it (see Prerequisites). Restart terminal if needed. |
| 401 Unauthorized | API key invalid or expired. |
| 400 Bad Request | Check prompt, size format, frame count rule (8n+1, ≤441). |
| 503 Service Unavailable | Service busy. Wait 30s, retry. |
| Video timeout (15 min) | Script prints the video_id — re-query later, the task is not abandoned. |
For full API behavior see references/api_docs.md.
Reference Files
references/api_docs.md— full Agnes API specificationreferences/prompt-guide.md— prompt structure for both image and videoreferences/parameters.md— every script parameter, with examples