frontend-vue

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

• Third-party content exposure detected (high risk: 0.70). SKILL.md:8-10 describes fetching and reading cached documentation content (via python3 scripts/cache_docs.py ... --permalink <permalink> and then reading __00.md/__01.md etc.), which is likely sourced from public web content at runtime (e.g., the “WebFetch URL” entries), so outsider-authored free text can be ingested into the LLM context.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill's runtime cache step ("python3 scripts/cache_docs.py ...") (SKILL.md:8) indicates it fetches documentation by permalink and the table lists external doc URLs such as https://jeesite.com/docs/vue-install-deploy/ (and other https://jeesite.com/docs/... pages) which are likely fetched at runtime and injected into agent responses, meaning remote content directly controls prompts.