kr-patent-claims-review
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-uploaded PDF and Word documents, which serves as a potential vector for indirect prompt injection attacks.
- Ingestion points: The agent reads user-provided patent application files via the
filestool and OCR processing in Step 1. - Boundary markers: While the skill uses a structured execution flow, it does not explicitly define delimiters to isolate raw file content from its analytical instructions.
- Capability inventory: The skill has access to file-writing tools (
files.write,files.append), OCR capabilities, and network-enabled search tools (patsnap_search,patsnap_fetch). - Sanitization: No explicit sanitization or filtering of the extracted document text is mentioned before processing.
- [EXTERNAL_DOWNLOADS]: The skill interacts with external services to retrieve patent data and prior art documentation.
- Details: It calls vendor-specific tools
patsnap_searchandpatsnap_fetchto query global patent databases. - Context: These resources are hosted on the vendor's official infrastructure (e.g.,
open.zhihuiya.com) and are essential for the skill's stated purpose of patent validity assessment.
Audit Metadata