nsfw-content-generation

Warn

Audited by Gen Agent Trust Hub on Apr 2, 2026

Risk Level: MEDIUMPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill explicitly instructs the agent to disable safety filters by setting the "enable_safety_checker" parameter to "false". This is designed to bypass standard content guidelines and safety constraints to facilitate the generation of adult-themed content.\n- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted user data.\n
  • Ingestion points: The "message" and "image_urls" parameters in "SKILL.md".\n
  • Boundary markers: None; user-provided strings and URLs are interpolated directly into API request bodies without delimiters.\n
  • Capability inventory: Performs network operations via "curl" to the vendor API at "https://sense.eachlabs.run".\n
  • Sanitization: No input sanitization or validation logic is present in the skill instructions.\n- [EXTERNAL_DOWNLOADS]: The skill facilitates network communication with the "eachlabs.run" domain for content generation and references documentation on "eachlabs.ai". It also allows for the ingestion of remote images through the "image_urls" array, which involves downloading data from external, user-controlled URLs.\n- [COMMAND_EXECUTION]: The documentation provides multiple shell command examples using "curl". These examples demonstrate how to make network requests and utilize environment variables for authentication.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 2, 2026, 02:04 AM