skills/nexscope-ai/nexscope-ecommerce-skills/ecommerce.tiktok-livestream-analytics/Gen Agent Trust Hub
ecommerce.tiktok-livestream-analytics
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external TikTok livestream leaderboards, creating a surface for potential adversarial content. \n
- Ingestion points: API responses from the Nexscope proxy are processed in
scripts/livestream_rank.pyandscripts/livestream_detail.py. \n - Boundary markers: The instructions do not provide delimiters or warnings for the agent to ignore instructions that might be embedded in the fetched TikTok data. \n
- Capability inventory: The skill has capabilities to write files to the local disk and perform network requests via the proxy. \n
- Sanitization: The skill does not implement specific sanitization or filtering for ingested strings such as livestream titles before they are presented to the agent. \n- [DATA_EXFILTRATION]: The scripts attempt to write session data to the user's home directory as a fallback location. \n
- Evidence: The
_nexscope_rootfunction in both Python scripts includesos.path.expanduser('~')in its search for a writable root directory to store JSON responses.
Audit Metadata