The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by processing untrusted user prompts and existing source code to generate and write new files to the filesystem. Ingestion points: User-provided natural language prompts requesting new feature implementations and existing project files read for structural context. Boundary markers: None identified; the skill instructions do not mandate delimiters or explicit warnings to the agent to ignore instructions embedded in the user input. Capability inventory: The skill facilitates reading existing code and writing multiple new files (controllers, services, entities, DTOs) across the 'src/' directory. Sanitization: No validation or sanitization logic is present to inspect user-provided prompts for malicious instructions before code generation occurs.
[SAFE]: The skill actively encourages security best practices by requiring the registration of global 'ValidationPipe' instances for input validation and mandating the use of explicit response DTOs with the 'class-transformer' library to prevent unintended data exposure.

nestjs-implementation