vss-search-index

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

• Third-party content exposure detected (medium risk: 0.30). In SKILL.md the outsider-provided natural-language query and optional tags/timeFilter are sent verbatim at runtime via POST /manager/search/query, and the workflow then ingests and parses the returned search results object for ranking/timestamps.

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

• Potentially malicious external URL detected (high risk: 1.00). The bootstrap script (scripts/vss-bootstrap.sh) may perform a runtime git clone of the repository https://github.com/open-edge-platform/edge-ai-libraries.git and then use/run the fetched application sources (it clones via git at runtime and the skill then cd's into the resolved app root to run commands), so remote code is fetched and relied upon during skill execution.