patent-layout-analysis
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements a structured workflow for patent analysis, including market research and competitive landscape evaluation. All external references target the author's official domain (zhihuiya.com) or well-known patent jurisdictions.
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests untrusted data from web searches and patent databases.
- Ingestion points: Market scale and trend search (Step 3), Global patent retrieval (Step 4), and leading company patent data (Step 5) in SKILL.md.
- Boundary markers: The instructions mandate a highly structured output format (Section 11) with specific modules, which provides implicit structure for the ingested data.
- Capability inventory: The agent is authorized to write session files (
@session/reports/) and perform network searches via MCP tools. - Sanitization: The skill does not explicitly define sanitization logic for HTML generation, but it specifies a rigid modular layout that limits the influence of external content on the overall agent logic.
Audit Metadata