Here's a comprehensive prompt you can use to evaluate a codi...

Prompt

Here's a comprehensive prompt you can use to evaluate a coding-generation model (LLM, coding agent, or autonomous software engineer). It is designed to test much more than code generation—it evaluates requirements analysis, architecture, implementation quality, testing, documentation, security, maintainability, and engineering judgment. --- # Coding Agent Evaluation Prompt You are an experienced Staff Software Engineer and Software Architect. Your objective is **not just to generate code**, but to demonstrate professional software engineering from requirements through deployment. Treat this as if you were joining an existing engineering team working on a production system. --- ## Project Build a **Task Management REST API** similar to a lightweight Jira/Trello backend. Users can: - Create projects - Create tasks - Assign tasks - Update task status - Search tasks - Add comments - View activity history --- # Functional Requirements ## Authentication Implement JWT authentication. Users must: - Register - Login - Refresh token Roles - Admin - Manager - User Permissions Admin - Everything Manager - Manage users inside their projects User - Only modify tasks assigned to them --- ## Projects CRUD Fields - Id - Name - Description - Created Date - Owner --- ## Tasks CRUD Fields - Title - Description - Priority - Status - Due Date - Story Points - Labels Relationships - belongs to Project - assigned User - comments - activity history Status - Todo - In Progress - Blocked - Done Priority - Low - Medium - High - Critical --- ## Search Support filtering - Status - Priority - Project - Assigned User - Due Date - Labels Sorting Pagination Full-text search --- ## Comments Users can comment on tasks. Comments maintain history. --- ## Activity Log Automatically log Task Created Task Updated Assignment Changed Status Changed Comment Added Deleted --- # Non-functional Requirements RESTful API OpenAPI documentation JWT Role-based authorization Optimistic concurrency Validation Error handling Logging Rate limiting Caching Repository Pattern CQRS Dependency Injection Configuration Unit Tests Integration Tests Docker Docker Compose Health Checks Observability Metrics Structured logging --- # Technical Stack .NET 10 ASP.NET Core Minimal APIs OR Controllers Entity Framework Core PostgreSQL Redis MediatR FluentValidation Serilog xUnit Testcontainers Swagger --- # Deliverables Generate --- ## 1 Requirements Analysis Identify - assumptions - ambiguities - risks - missing requirements --- ## 2 Domain Model Design Entities Value Objects Enums Relationships Aggregates --- ## 3 Architecture Explain - Clean Architecture - CQRS - Folder structure - Dependency graph --- ## 4 Database Generate ER Diagram Indexes Constraints Migration strategy --- ## 5 API Design Produce REST endpoints Request models Response models Validation HTTP status codes Example payloads --- ## 6 Security Review Discuss Authentication Authorization OWASP Top 10 SQL Injection Mass Assignment IDOR CSRF Secrets management Rate limiting --- ## 7 Implementation Generate production-ready code. Requirements No TODOs No placeholders No pseudocode Use best practices. --- ## 8 Testing Generate Unit tests Integration tests Authorization tests Repository tests Validation tests API tests Edge cases Failure cases --- ## 9 Performance Discuss Indexes Caching N+1 queries Connection pooling Pagination Async Memory Scalability --- ## 10 Observability Implement Structured logging Tracing Metrics Health checks Correlation IDs --- ## 11 CI/CD Generate GitHub Actions Dockerfile docker-compose Migration pipeline Deployment pipeline --- ## 12 Documentation Produce README Architecture Decision Records (ADRs) Setup instructions Developer guide API documentation --- # Refactoring Before producing the final solution: 1. Review the generated architecture. 2. Identify weaknesses. 3. Refactor where appropriate. 4. Eliminate duplication. 5. Simplify complexity. 6. Improve naming. 7. Improve extensibility. 8. Ensure SOLID principles. 9. Ensure Clean Architecture boundaries. 10. Verify dependency inversion. --- # Self Evaluation After implementation provide a brutally honest review. Evaluate Architecture Code quality Readability Maintainability Security Performance Scalability Testing Developer experience Production readiness For every weakness - explain why - estimate severity - propose improvements --- # Output Format Produce the following sections in order: 1. Executive Summary 2. Requirements Analysis 3. Architecture 4. Domain Model 5. Database Design 6. API Specification 7. Security Review 8. Implementation 9. Testing Strategy 10. Performance Analysis 11. Observability 12. CI/CD 13. Documentation 14. Self Review 15. Improvement Roadmap --- # Evaluation Criteria The solution will be scored on: | Category | Weight | |----------|---------:| | Requirements understanding | 10% | | Architecture quality | 15% | | Code correctness | 15% | | Security | 10% | | Testing completeness | 10% | | Performance | 10% | | Maintainability | 10% | | Production readiness | 10% | | Documentation | 5% | | Self-critique and engineering judgment | 5% | A top-performing coding model should demonstrate not only the ability to generate working code, but also the reasoning, design trade-offs, quality assurance, and operational considerations expected from a senior software engineer building production-grade systems.