restaurant-local-buzz

Pass

Audited by Gen Agent Trust Hub on Jul 2, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill retrieves and processes untrusted external data from customer reviews and social media comments, which presents a surface for indirect prompt injection.
  • Ingestion points: The skill is designed to read live data from TikTok comments (tiktok/videos/{id}/comments) and review text samples from Google Maps and search listings (SKILL.md).
  • Boundary markers: The instructions do not specify any delimiters or safety prompts to ensure the agent ignores malicious instructions embedded within the external data.
  • Capability inventory: The skill's functions are limited to data analysis, scoring, and report generation. It does not have access to sensitive local files, system commands, or write operations.
  • Sanitization: No sanitization or filtering logic is described for the content retrieved from external sources.
  • [EXTERNAL_DOWNLOADS]: The skill references external vendor resources for setup and data access.
  • The documentation instructs users to install a package using npx skills add unifapi-agent/agents (README.md).
  • The skill establishes a connection to the vendor's MCP server at https://mcp.unifapi.com (README.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 2, 2026, 10:29 AM
Security Audit — agent-trust-hub — restaurant-local-buzz