Framework Architecture
ECC is built on five interconnected components that work together to create a powerful development environment:Agents
Specialized subagents for delegated tasks with limited scope
Skills
Workflow definitions and domain knowledge libraries
Commands
Slash commands for quick execution of common workflows
Hooks
Event-driven automations that fire on tool executions
Rules
Always-follow guidelines for code quality and standards
Core Principles
ECC is designed around five fundamental principles:1. Agent-First Development
Delegate domain-specific tasks to specialized agents rather than handling everything in the main conversation. This provides:- Focused expertise - Each agent is optimized for specific tasks
- Better context management - Agents work with limited scope
- Parallel execution - Independent agents can run simultaneously
2. Test-Driven Development
All code must be developed test-first with comprehensive coverage:- Write tests before implementation (RED → GREEN → REFACTOR)
- 80%+ coverage requirement across unit, integration, and E2E tests
- Edge cases and error scenarios must be tested
3. Security-First Approach
Security is never compromised:- No hardcoded secrets (API keys, passwords, tokens)
- All user inputs validated at system boundaries
- SQL injection and XSS prevention enforced
- Authentication and authorization verified before commits
4. Immutability
Always create new objects, never mutate existing ones:5. Plan Before Execute
Complex features require planning before implementation:- Use the planner agent for architectural changes
- Break down features into manageable phases
- Identify dependencies and risks upfront
- Enable incremental testing and delivery
Project Structure
How Components Work Together
Success Metrics
ECC measures success through five key metrics:- ✅ All tests pass with 80%+ coverage
- ✅ No security vulnerabilities detected
- ✅ Code is readable and maintainable
- ✅ Performance meets requirements
- ✅ User requirements are fully met
Cross-Platform Support
ECC works across multiple AI coding tools:| Platform | Agents | Commands | Skills | Hooks | Rules |
|---|---|---|---|---|---|
| Claude Code | 13 | 33 | 50+ | 8 events | 29 |
| Cursor IDE | Shared | Shared | Shared | 15 events | 29 |
| Codex CLI | Shared | Instruction-based | 16 | Not yet | Instruction-based |
| OpenCode | 12 | 24 | 37 | 11 events | 13 |
All platforms share the same
AGENTS.md file at the root, ensuring consistent behavior across tools.Next Steps
Understand Agents
Learn how specialized agents handle delegated tasks
Explore Skills
Discover workflow definitions and domain knowledge
Use Commands
Quick reference for slash commands
Configure Hooks
Set up event-driven automations