菜单
Creative idea advisor. Brainstorm frameworks, product archetypes, industry data, and monetization models. Zero-token data retrieval via script.
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
Auto-generate concise, visually stunning README.md with project auto-scanning, badges, and 3 style modes.
Generate unit/integration tests with smart edge case analysis (Zero Token Skeleton + Smart Mode).
Persistent team profiles — carry your coding style, rules, and bug-fix knowledge across projects. 3-tier memory (Hot/Warm/Cold) with TF-IDF search.
Deploy local websites to the internet instantly via Cloudflare Tunnel. Zero hosting, zero domain needed.
Suggests suitable tech stack based on project requirements using local database.
Analyzes market trends, hot features, and user expectations.
| name | meta-thinker |
| description | Creative idea advisor. Brainstorm frameworks, product archetypes, industry data, and monetization models. Zero-token data retrieval via script. |
Provides rich data for brainstorming and idea development. Instead of consuming tokens to "think", the agent uses a script to retrieve existing data.
| File | Content | Entries |
|---|---|---|
data/brainstorm_frameworks.json | 15 creative frameworks (SCAMPER, Lean Canvas, TRIZ, Disney...) | 15 |
data/product_archetypes.json | 25 product templates + examples | 25 |
data/industry_database.json | 45 industries + pain points + opportunities | 45 |
data/feature_ideas.json | 300+ feature ideas by 23 categories | 300+ |
data/monetization_models.json | 16 monetization models + examples | 16 |
data/platform_guide.json | 6 platforms + pros/cons + tech stacks | 6 |
# Find ideas by domain
python scripts/idea_engine.py --domain "food-delivery" --query "healthy"
# Get features for a product type
python scripts/idea_engine.py --archetype "marketplace" --platform "mobile"
# Find suitable monetization
python scripts/idea_engine.py --monetization --domain "education"
# Brainstorm with a framework
python scripts/idea_engine.py --framework "lean-canvas" --context "fitness app"
The script returns compact JSON, the agent just needs to read it and present it to the user. Saves ~90% tokens compared to self-brainstorming without data.