enterprise-patent-report
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from multiple sources to construct its final output. Without explicit sanitization or strict boundary markers, malicious instructions embedded in these inputs could influence the agent's behavior during the report construction process.
- Ingestion points: SKILL.md Step 0 collects enterprise names, industry keywords, and specifically requires the user to upload six PowerPoint attachments (e.g., '专利布局规划解决方案') which are parsed to populate the report.
- Boundary markers: No specific delimiters or 'ignore embedded instructions' warnings are instructed for the agent when processing the content of the user-provided PPT files or metadata.
- Capability inventory: The skill utilizes MCP tools to query external patent databases and writes generated HTML content to files in the
@session/directory. - Sanitization: There are no instructions for escaping or validatng user-provided strings or file content before they are interpolated into the final HTML document.
- [DYNAMIC_EXECUTION]: The skill dynamically generates and writes HTML and CSS files to local session storage. While necessary for its purpose as a report generator, the assembly of these files from external data (such as patent abstracts or user-uploaded presentation content) creates a surface for potential cross-site scripting (XSS) if malicious data is injected into the source documents.
Audit Metadata