The Agent Skills Directory

[SAFE]: The skill provides production-ready templates and best practices for FastAPI, including proper use of async def for all handlers and services.
[EXTERNAL_DOWNLOADS]: Mentions github.com/wshobson/agents as the source of its base patterns. This is documented as attribution and does not instruct the agent to execute remote code at runtime.
[COMMAND_EXECUTION]: Provides instructions for managing dependencies and running the server using uv and granian. These are legitimate developer tools.
[SAFE]: Includes templates for configuration using pydantic-settings and environment variables. This follows standard security practices by avoiding hardcoded credentials in the codebase.
[PROMPT_INJECTION]: Defines endpoints for processing external data, such as AI chat streaming. While this creates a potential attack surface for indirect prompt injection, it is a standard functional requirement for the described use case.
Ingestion points: Data enters via Pydantic schemas in request bodies and query parameters across the api/v1/ routes.
Boundary markers: Not present in the generic templates.
Capability inventory: Includes the ability to run shell commands via uv and access a PostgreSQL database via psycopg.
Sanitization: Uses Pydantic for structural validation, though specific content filtering is left to the developer.

fastapi-guideline