chrome-extension-consultant

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

• Third-party content exposure detected (high risk: 1.00). The skill's reference files explicitly show runtime ingestion of untrusted web content—e.g., references/extension-architecture.md defines a FETCH_EXTERNAL handler that does fetch(payload.url), references/extension-architecture.md and references/extension-architecture.md (Content Script Patterns) inject scripts and read window.someData from web pages, and references/manifest-v3-guide.md lists broad host_permissions (including "<all_urls>"), so the agent/extension is expected to fetch and interpret arbitrary third-party page/API content that could influence behavior.