Agent Skills
› SourceWeft/SourceWeft
› image-generate
image-generate
GitHub基于文本提示生成持久化图像资源,适用于创建位图、海报、产品模型等视觉内容。
Trigger Scenarios
用户请求生成或绘制图像
需要制作产品样机或概念艺术
Install
npx skills add SourceWeft/SourceWeft --skill image-generate -g -y
SKILL.md
Frontmatter
{
"name": "image-generate",
"description": "Generate persisted image artifacts from text prompts. Use when the user asks to create, render, draw, illustrate, design, or generate a bitmap image, poster visual, product mockup, concept art, or other standalone raster visual artifact through SourceWeft image generation."
}
Image Generate
Use SourceWeft's built-in generate_image tool to create persisted image artifacts.
Tool
Call generate_image with a concise visual prompt. The callable arguments are prompt and optional title; aspect ratio, quality, and style are configured by SourceWeft options.
Workflow
- Treat the user request as an image brief.
- Preserve the requested subject, style, composition, visible text, and constraints.
- Expand vague requests only enough to make the image prompt clear.
- Call
generate_imageonce for each distinct image.
Boundaries
- Do not use a sandbox or filesystem workflow for this skill.
- Use a code-native or vector workflow when the user explicitly asks for SVG, HTML/CSS/canvas, charts, technical diagrams, or changes to an existing logo/icon system.
Output
- Return a generated image artifact, not only a text description.
- Do not return raw file paths, raw artifact URLs, or image markdown; SourceWeft displays the artifact.
- If generation succeeds, respond with at most one short sentence.
- If
generate_imageis unavailable or fails, report the failure briefly.
Version History
- 66a6a34 Current 2026-11-07 08:57


