跳到正文
TakoAPI
返回技能列表

creative-toolkit

由 jau123 提供

Coding Agents & IDEs
925下载量由 clawskills.sh 提供
1星标
0浏览量
0 个赞登录后即可为这个技能点赞
—暂无评分
你的评分

简介

Generate images from text with multi-provider routing — supports Nanobanana Pro, GPT Image, Seedream, and local.

描述

该技能的功能

A unified MCP server for AI image generation that routes requests across multiple providers including MeiGen Cloud, GPT Image, and local ComfyUI. Includes a searchable gallery of 1,300+ curated prompts and a prompt enhancement tool that expands brief ideas into detailed, style-aware descriptions.

Routing across multiple providers from a single interface means switching backends requires only a config change, not a code rewrite.

何时使用

  • Expanding a short concept into a detailed image prompt before generating
  • Searching a gallery of curated prompts for creative starting points
  • Running image generation through a local ComfyUI instance without external APIs
  • Generating logo mockups using an existing image as a style reference
  • Managing ComfyUI workflow templates and adjusting sampler settings without editing JSON

示例工作流

输入

User asks: generate a futuristic city skyline at night, realistic style

智能体
  1. 1

    Call enhance_prompt with brief='futuristic city skyline at night' and style='realistic' to expand the idea into a detailed prompt with lighting and composition details

  2. 2

    Review the enhanced prompt returned by the tool

  3. 3

    Call generate_image with the enhanced prompt and aspectRatio='16:9'

  4. 4

    Present the exact image URL and saved file path from the tool response to the user

输出

Image URL and local file path for the generated image

前置要求

  • MEIGEN_API_TOKEN from meigen.ai (for MeiGen Cloud provider)
  • API key from Together AI, Fireworks AI, or another OpenAI-compatible provider (if using that option)
  • Running local ComfyUI instance (if using ComfyUI provider)

安装

clawhub install jau123/creative-toolkit

运行 CLI 命令,或把这段提示词粘贴到 Claude Code、Cursor、Codex 或你在用的任意编程智能体里。

链接