AGENT LAB: SKILLS

prompt-engineering

Fail

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • REMOTE_CODE_EXECUTION (HIGH): The skill instructs the agent/user to execute curl -fsSL https://cli.inference.sh | sh. This pattern is highly insecure as it pipes unverified remote content directly into a shell interpreter, allowing the server at cli.inference.sh to execute any command on the user's machine.
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill depends on a CLI tool (infsh) hosted on a non-standard, third-party domain (inference.sh) that is not included in the trusted source whitelist. There is no integrity verification (e.g., checksums) for this download.
  • COMMAND_EXECUTION (LOW): The allowed-tools configuration Bash(infsh *) provides the agent with broad permission to execute any subcommand of the infsh tool, which could be exploited if the tool has undocumented or dangerous capabilities.
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