The Agent Skills Directory

[METADATA_POISONING]: The skill exhibits a discrepancy between the registered author ('dennislee928') and the maintainer details provided in the documentation ('Jezweb | jeremy@jezweb.net'). Additionally, the documentation includes several references to 'future' dates in 2026 (e.g., 'Last verified: 2026-01-21'), which is misleading regarding the actual timeline of maintenance and the skill's origin.
[INDIRECT_PROMPT_INJECTION]: The skill's API templates in src/items/router.py and src/auth/router.py ingest untrusted user input via Pydantic schemas. Although the templates employ standard Pydantic validation (e.g., min_length, gt=0), they do not include boundary markers or instructions to downstream language models to disregard potential prompt injection attacks embedded in user-supplied strings. The skill grants the capability to persist this data into a database as seen in the SQLAlchemy configuration in src/database.py and src/items/models.py.

fastapi