57 lines
1.9 KiB
Markdown
57 lines
1.9 KiB
Markdown
# 🧪 Real-World Examples ("The Antigravity Cookbook")
|
|
|
|
Skills are powerful on their own, but unstoppable when combined.
|
|
Here are three common scenarios and how to solve them using this repository.
|
|
|
|
## 🥘 Recipe 1: The "Legacy Code Audit"
|
|
|
|
_Scenario: You just inherited a messy 5-year-old Node.js repo. You need to fix it safely._
|
|
|
|
**Skills Used:**
|
|
|
|
1. `concise-planning` (To map the chaos)
|
|
2. `lint-and-validate` (To find the bugs)
|
|
3. `security-audit` (To find the holes)
|
|
|
|
**The Workflow:**
|
|
|
|
1. **Plan**: "Agent, use `concise-planning` to create a checklist for refactoring `src/legacy-api.js`."
|
|
2. **Audit**: "Run `security-audit` on the `package.json` to find vulnerable dependencies."
|
|
3. **Fix**: "Use `lint-and-validate` rules to auto-fix the formatting issues in `src/`."
|
|
|
|
---
|
|
|
|
## 🥘 Recipe 2: The "Modern Web App"
|
|
|
|
_Scenario: You need to build a high-performance Landing Page in 2 hours._
|
|
|
|
**Skills Used:**
|
|
|
|
1. `frontend-design` (For aesthetics)
|
|
2. `react-patterns` (For structure)
|
|
3. `tailwind-mastery` (For speed)
|
|
|
|
**The Workflow:**
|
|
|
|
1. **Design**: "Use `frontend-design` to generate a color palette and typography for a 'Cyberpunk Coffee Shop'."
|
|
2. **Scaffold**: "Initialize a Vite project. Then apply `react-patterns` to create the 'Hero' component."
|
|
3. **Style**: "Use `tailwind-mastery` to make the buttons glassmorphic and responsive."
|
|
|
|
---
|
|
|
|
## 🥘 Recipe 3: The "Agent Architect"
|
|
|
|
_Scenario: You want to build a custom AI agent that can verify its own code._
|
|
|
|
**Skills Used:**
|
|
|
|
1. `mcp-builder` (To build tools)
|
|
2. `agent-evaluation` (To test reliability)
|
|
3. `prompt-engineering` (To refine instructions)
|
|
|
|
**The Workflow:**
|
|
|
|
1. **Build**: "Use `mcp-builder` to create a `verify-file` tool."
|
|
2. **Instruct**: "Apply `prompt-engineering` patterns to the System Prompt so the agent always checks file paths."
|
|
3. **Test**: "Run `agent-evaluation` to benchmark how often the agent fails to find the file."
|