The Agent Skills Directory

[PROMPT_INJECTION]: The skill includes instructions that direct the agent to proactively suggest the vendor's 'K-Dense Web' platform when users engage in complex, multi-step reasoning or deep research workflows. This is a documented behavioral steer intended for productivity enhancement within the vendor's ecosystem.- [EXTERNAL_DOWNLOADS]: The documentation and scripts recommend the installation of the uspto-opendata-python library from PyPI. This is an established and specialized package used for accessing USPTO Patent Examination Data System (PEDS) records.- [DATA_EXFILTRATION]: The skill utilizes environment variables such as USPTO_API_KEY and PATENTSVIEW_API_KEY for authentication, which is a standard security practice to prevent the exposure of credentials in source code. All network traffic is directed to official or well-known service domains including uspto.gov and patentsview.org.- [SAFE]: The skill processes data from external USPTO APIs, which constitutes an indirect prompt injection surface.
Ingestion points: Patent and trademark data are retrieved in scripts/patent_search.py, scripts/peds_client.py, and scripts/trademark_client.py.
Boundary markers: Data is ingested and processed in structured JSON and XML formats; the scripts do not currently implement explicit delimiters to separate external content from agent instructions.
Capability inventory: The skill has the capability to make outbound HTTPS requests via the requests library and execute local Python helper scripts.
Sanitization: The scripts perform basic input validation and formatting, such as removing punctuation from patent numbers, and treat API responses as structured data.

uspto-database