linkfox-kalodata-tiktok-creator

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPERSISTENCEDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes external content from TikTok creator profiles, including the creator_bio field, via the scripts/kalodata_creator_detail.py script. This represents an ingestion point for untrusted data that could influence agent behavior.
  • Ingestion points: External creator metadata fetched from the Kalodata API in scripts/kalodata_creator_detail.py.
  • Boundary markers: The prompt instructions do not provide specific delimiters or 'ignore' instructions for the agent when processing profile data.
  • Capability inventory: The skill possesses network access to vendor APIs and file-writing capabilities for logging and caching.
  • Sanitization: The scripts do not perform sanitization of the bio text or other descriptive fields before returning them to the agent context.
  • [DYNAMIC_EXECUTION]: The scripts kalodata_creator_detail.py and kalodata_creator_search.py dynamically modify the Python sys.path at runtime to import shared modules from a relative parent directory (../../_shared). This is a standard practice for internal library management within the vendor's project structure.
  • [PERSISTENCE]: The references/onboarding.md file provides instructions for users to manually add the LINKFOX_AGENT_API_KEY to their shell profiles (e.g., .zshrc, .bashrc) or Windows environment variables. This establishes persistent access for the authentication credentials.
  • [EXTERNAL_DOWNLOADS]: The skill communicates with vendor-controlled domains including tool-gateway.linkfox.com, api.linkfox.com, and agent-api.linkfox.com. These resources are used for data access, login services, and user management consistent with the vendor's infrastructure.
  • [COMMAND_EXECUTION]: The onboarding documentation guides the user to manually execute shell commands to configure their environment variables. The skill scripts themselves do not automatically execute these system-level commands.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 07:11 AM