ads-tiktok
Pass
Audited by Gen Agent Trust Hub on Sep 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data from
tiktok-data.jsonand a community MCP server. This data includes campaign names, ad captions, and creative text which are potentially attacker-controlled. - Ingestion points: Data is pulled from the
AdsMCP/tiktok-ads-mcp-serverand thetiktok-data.jsonfile. - Boundary markers: The instructions do not define clear delimiters or "ignore embedded instructions" warnings for the agent when processing the ad data.
- Capability inventory: The skill performs complex NLP analysis and generates reports in both Markdown and JSON formats based on the ingested data.
- Sanitization: There is no evidence of sanitization or validation of the text content retrieved from the TikTok Ads platform.
- [EXTERNAL_DOWNLOADS]: The skill instructions direct the user to set up a community-maintained MCP server (
AdsMCP/tiktok-ads-mcp-server) and utilize local scripts (scripts/api/tiktok_fetch.py). These represent dependencies on external code that is not part of a verified or trusted vendor list, requiring the user to trust the source of the community tools.
Audit Metadata