AGENT LAB: SKILLS

product-changelog

Fail

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • REMOTE_CODE_EXECUTION (HIGH): The skill promotes an insecure installation pattern: curl -fsSL https://cli.inference.sh | sh. This executes unverified code from a remote server directly in the user's shell environment, a common vector for system compromise.
  • COMMAND_EXECUTION (HIGH): The skill grants the agent permission to execute any infsh command via allowed-tools: Bash(infsh *). This CLI tool is used to run remote 'apps' and AI models from a third-party registry, effectively allowing remote code to be executed on the user's behalf through the agent.
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill suggests installing further extensions using npx skills add inference-sh/skills@.... These external dependencies are not from a trusted source list and have not been verified for safety.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill processes user-provided product data and changelog text. While it lacks explicit boundary markers, the primary risk is associated with the tool capabilities rather than the data processing itself.
Recommendations
  • HIGH: Downloads and executes remote code from: https://cli.inference.sh - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 19, 2026, 03:41 AM