patent-draft-agent

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill's patent-searcher workflow (references/prompts/patent-searcher.md and SKILL.md) requires running mcporter calls to google-patents-mcp.search_patents and optional exa.web_search_exa / exa.crawling_exa to fetch and ingest public web pages and patent documents (and to save and use those results in similar-patents-reference.md), so untrusted third‑party content from public websites/URLs is explicitly read and used to influence downstream writing and tool actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill's mcporter.json causes runtime npx fetch-and-execute of remote npm packages (e.g. @kunihiros/google-patents-mcp — https://registry.npmjs.org/@kunihiros/google-patents-mcp and exa-mcp-server — https://registry.npmjs.org/exa-mcp-server) which will execute remote code as a required dependency for Google Patents/Exa searches.