image-generator

Files (1)

• SKILL.md

SKILL.md (first lines)

# Image Generation Agent

You create images from natural language descriptions.

## When to use this skill
- The user asks to generate, draw, create, or make an image / picture / illustration / artwork.
- The request contains visual descriptions (e.g. “a cyberpunk city at night”, “portrait of a fox in watercolor”).

## Fallback chain (server `/api/image`)
1. **Hugging Face FLUX.1-schnell** (primary)
2. **Prexzy** endpoints (genimage → txt2img → dalle → aiwriter) — backup

Prefer `PrexzyAPI.generateImage({ prompt, size }, { loadingEl })` from the client.
That helper routes through `/api/image`, consumes the `image` quota once, and returns `{ url, source }`.

## Parameters
| Name    | Required | Description                                      | Example                  |
|---------|----------|--------------------------------------------------|--------------------------|
| prompt  | yes      | Full detailed description of the desired image   | "a red fox in a snowy forest, cinematic lighting" |
| size    | no       | Output size                                      | "1024x1024", "1024x1792", "1792x1024" |

## Instructions for the agent
- Always pass the complete user description as the `prompt`.
- If the user specifies orientation (portrait / landscape / square), map it to the correct `size`.
- After receiving the result, return the image URL (or base64 data) so the UI can display it.
- If the call fails, return a clear error message and the remaining quota for the `image` feature.

## Example
User: “Make a cinematic image of a lone samurai standing under cherry blossoms at dusk”
→ Call `PrexzyAPI.generateImage` with prompt = the full sentence above.

From awesomejoses11-code/Canton-Node, CC0-1.0.

More from awesomejoses11-code/Canton-Node