AGENTS.md Sections

Sylva generates a standardized AGENTS.md with 17 sections. Each section provides specific context that AI coding tools use to generate more accurate, project-aware code.

Section Reference

1. Project Overview

High-level description of what the project does, its sub-services, and primary purpose.

2. Agent Persona

Optional persona description for AI assistants working on this project (e.g., "You are a backend engineer specializing in FastAPI").

3. Tech Stack

Every language, framework, library, database, and external API used. Includes versions when identifiable from dependency manifests.

4. Architecture

Directory layout with module responsibilities. Includes an ASCII architecture diagram showing relationships between sub-services and external integrations.

Example output:

myshabeauty/
├── frontend/
│   ├── components/
│   │   ├── HomePage.jsx
│   │   ├── ProductDetailPage.jsx
│   │   └── CartDrawer.jsx
│   └── contexts/
│       ├── AuthContext.js
│       └── CartContext.js
├── backend/
│   ├── server.py
│   └── wix_service.py
└── tests/
    └── test_*.py

5. Code Style

Naming conventions, formatting rules, import ordering, type-hinting practices, and linter configurations. Separated by stack when applicable (e.g., frontend vs backend).

6. Anti-Patterns & Restrictions

Explicit rules about what NOT to do. Examples:

  • "NEVER hardcode discount/badge UI"
  • "NEVER commit API tokens"
  • "NEVER hardcode CORS origins"

7. Database & State Management

Database technology, ORM/ODM patterns, and state management strategies (e.g., React Context, Redux, Vuex).

8. Error Handling & Logging

Error handling patterns, logging libraries, and structured logging conventions.

9. Testing Commands

Exact commands to run tests:

pytest
npm test
npm run test:e2e

10. Testing Guidelines

Test frameworks, naming conventions, file placement rules, and mocking strategies.

11. Security & Compliance

Authentication patterns, secret management, CORS configuration, and compliance requirements.

12. Dependencies & Environment

How to install dependencies, manage environments, and configure build tools.

13. PR & Git Rules

Branching strategy, commit message conventions, and CI requirements.

14. Documentation Standards

How documentation should be written and maintained.

15. Common Patterns

Frequently used code patterns in the project (e.g., "Always use Context for shared state", "Debounce network calls").

16. Agent Workflow / SOP

Step-by-step standard operating procedure for AI agents:

  1. Classify the change (frontend, backend, tests)
  2. Check existing patterns
  3. Implement following conventions
  4. Add tests
  5. Validate deployment impact

17. Few-Shot Examples

Concrete "good" and "bad" code examples showing project-specific patterns:

// Good: Using Context for Auth
import { createContext, useContext } from 'react';
const AuthContext = createContext();
export const useAuth = () => useContext(AuthContext);

// Bad: Hardcoding values
const DiscountLabel = "10% OFF"; // Should derive from backend

results matching ""

    No results matching ""