The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes a file named patent-assistant_audit_result_v1.json that contains pre-computed security and stability verdicts (e.g., "security": "PASS", "deployable": true). This represents a self-referential indirect injection pattern intended to influence the conclusions of security analysis tools. • Ingestion points: patent-assistant_audit_result_v1.json • Capability inventory: scripts/patent_search.py (Network GET requests), scripts/generate_disclosure.py (File I/O) • Boundary markers: None • Sanitization: None
[INDIRECT_PROMPT_INJECTION]: The skill processes data from external search results via scripts/patent_search.py. The ingestion of untrusted content from the web presents an attack surface where maliciously crafted patent titles or abstracts could influence agent behavior. • Ingestion points: scripts/patent_search.py (via search engine responses) • Capability inventory: scripts/patent_search.py (Network GET requests), scripts/generate_disclosure.py (File I/O) • Boundary markers: None • Sanitization: None
[EXTERNAL_DOWNLOADS]: The scripts/patent_search.py script performs network operations to retrieve patent data from well-known intellectual property and academic search platforms, including Google Patents, Lens.org, Innojoy, Baidu Academic, Espacenet, and CNIPA. These operations are consistent with the skill's primary function.
[COMMAND_EXECUTION]: The skill instructions provide examples for the user to execute included Python scripts (scripts/generate_disclosure.py and scripts/patent_search.py) to process technical descriptions and perform searches.

patent-assistant