The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill contains hardcoded default credentials in connection strings and configuration files intended for development environments.\n
Evidence: Hardcoded password '123456' in connection strings within Part 7.1 and Part 14.\n
Evidence: 'POSTGRES_PASSWORD: "123456"' defined in the docker-compose.yml configuration in Part 14.\n- [INDIRECT_PROMPT_INJECTION]: The skill describes an architecture for a backend system that processes untrusted external data, which presents an attack surface for indirect prompt injection.\n
Ingestion points: The system handles external data through API request DTOs (e.g., UserCreate, UserUpdate) and Excel file parsing using the calamine crate.\n
Boundary markers: The instructions recommend using the validator crate to enforce schema constraints on incoming data.\n
Capability inventory: The described system has capabilities for database manipulation (SeaORM), file system writes (Excel generation), and network operations (email sending via lettre).\n
Sanitization: Data is structured using serde and validated with the validator crate, but the ingestion of complex formats like Excel remains a vulnerability surface.\n- [DYNAMIC_EXECUTION]: The skill includes a code generation component which involves the programmatic creation of source code and modules.\n
Evidence: The skill mentions a codegen module in Part 2 and Part 15 intended for generating business logic, handlers, and entities from predefined templates.

rust