The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The setup instructions in SKILL.md include the command curl -LsSf https://astral.sh/uv/install.sh | sh. Piping a remote script directly to a shell allows for arbitrary code execution from a third-party domain without inspection or integrity pinning.\n- [CREDENTIALS_UNSAFE]: The configuration example in src/app/config.py provides hardcoded sensitive defaults: database_url with embedded credentials (user:pass) and a secret_key (change-me-in-production). These placeholders normalize insecure coding practices and can lead to compromised environments if not properly managed via environment variables.\n- [COMMAND_EXECUTION]: The skill relies on shell commands for project initialization, dependency management, and server execution using tools like uv, alembic, and uvicorn.\n- [EXTERNAL_DOWNLOADS]: The skill requires downloading external code from astral.sh and various third-party libraries from Python package registries.\n- [DATA_EXFILTRATION]: The recommended uvicorn commands bind the application to 0.0.0.0, potentially exposing the service to external networks. Combined with the CORSMiddleware wildcard origin (*), this creates a risk of unauthorized cross-origin data access.\n- [PROMPT_INJECTION]: The skill exposes an indirect prompt injection surface. Ingestion points: Untrusted user data enters via FastAPI endpoints in src/app/features/users/api.py. Boundary markers: None are present to delimit user content. Capability inventory: The skill includes capabilities to write to a database via UserRepository and execute a web server. Sanitization: Relies on Pydantic schema validation for data types, but lacks explicit sanitization for potential injection payloads.

python-fastapi