face-swap
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external image URLs and text prompts provided by users or retrieved from the web. This creates a surface for indirect prompt injection where adversarial instructions could be embedded in image metadata or text content. This risk is inherent to the multimodal nature of the face-swapping task.
- [COMMAND_EXECUTION]: The skill specifies the use of the Bash tool with curl to transmit data to the each::sense API endpoint at eachsense-agent.core.eachlabs.run. The provided examples are well-defined and strictly support the primary function of image processing and transmission.
- [NO_CODE]: The skill consists entirely of instructional markdown and YAML configuration within the SKILL.md file. It does not ship with any external scripts, binaries, or executable dependencies, reducing the direct attack surface on the host system.
Audit Metadata