map-competitive-patent-landscape-ip
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests patent data including titles and abstracts from external databases via Model Context Protocol (MCP) servers, which represents a potential surface for indirect prompt injection if malicious instructions are embedded in the records.
- Ingestion points: Patent search and briefing data retrieved via the
advanced_patent_searchandpatent_briefingMCP tools described inSKILL.md. - Boundary markers: Instructions mandate a strict separation of retrieved facts from strategic interpretations and require labeling forward-looking implications as hypotheses.
- Capability inventory: The skill instructions involve generating and writing an HTML report file to the local workspace.
- Sanitization: The instructions specify using semantic HTML5 and accessible tables for the output, relying on the agent's inherent safety guardrails for the processing of external patent text.
- [EXTERNAL_DOWNLOADS]: The skill documentation references official connection URLs and documentation for Patsnap's patent search infrastructure (
open.patsnap.com). These references are associated with the primary functionality of the skill as provided by the vendor.
Audit Metadata