菜单
Use Google Gemini for text generation, image analysis, video analysis, and image generation
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Scrape product reviews from any ecommerce site into standardized JSONL. Use when the user says "scrape reviews," "pull reviews," "get reviews from," "grab reviews," or provides a product/review page URL and wants review data.
Generates advertising creative briefs for direct-response marketing using three psychological frameworks (TEEP stages, Three Selves theory, Emotional Zones) with voice of customer integration. Use when the user asks for creative briefs, ad briefs, advertising concepts, static ads, or mentions TEEP, Trigger, Exploration, Evaluation, Purchase, self-targeting, Actual self, Ideal self, Ought self, emotional zones, valence, intensity, emotional journey, voice of customer, VoC, angle dumps, or distribution specs.
Critique copy or creative work against a chosen skill/framework. Use when the user says "critique this," "review this," "score this," "what's wrong with this," "give me feedback," or "how can I improve this."
Generates editorial-style image prompts that look like real content, not ads. Guides users through style selection and builds hyper-detailed prompts for Nano Banana Pro (Gemini). Use when the user mentions editorial images, editorial style images, native ad images, or wants to generate images that don't look like ads.
Writes research-driven listicle landing pages where each numbered point is a sales argument built from a unifying theme. Guides through 9 decision gates covering research, theme selection, headlines, feature/benefit mapping, argument building, drafting, review, and imagery. Use when creating listicles, writing conversion-focused numbered lists, or when the user mentions listicles, landing pages, ad landing pages, or numbered list articles.
Generates native advertising creative (headlines + image direction) using direct response psychology. Use when writing native ads, creating ad concepts, brainstorming ad headlines, or developing native creative for health, finance, ecom, or supplement campaigns. Triggers on "native ad," "write headlines," "ad creative," "native creative," or "ad concept."
| name | gemini-api |
| description | Use Google Gemini for text generation, image analysis, video analysis, and image generation |
| trigger | gemini, use gemini, analyze with gemini, gemini image, gemini video, generate image |
Use the Gemini API for tasks Claude can't do natively: analyzing videos, generating images, or running prompts through Gemini.
The Gemini API wrapper is at tools/gemini-api/gemini-api.js relative to the repo root.
First-time setup:
cd tools/gemini-api && npm install
Add your API key to tools/gemini-api/.env:
GEMINI_API_KEY=your_key_here
node tools/gemini-api/gemini-api.js "your prompt here"
node tools/gemini-api/gemini-api.js "describe what you see" --image /path/to/image.jpg
node tools/gemini-api/gemini-api.js "describe this video in detail including all text, speech, and visual elements" --video /path/to/video.mp4
Supported video formats: .mp4, .mov, .avi, .webm
node tools/gemini-api/gemini-api.js "a cat on a beach" --generate-image --output /path/to/output.png
node tools/gemini-api/gemini-api.js "product shot" --generate-image --count 3 --output-dir /path/to/dir
Add --json flag to any text/analysis command to get structured JSON back.
Add --model model-name to use a specific model. Defaults:
gemini-2.5-progemini-2.0-flash-exp-image-generationGEMINI_API_KEY is set in tools/gemini-api/.env before use