NYC

logo-design-guide

Fail

Audited by Gen Agent Trust Hub on Feb 18, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • Unverifiable Dependencies & Remote Code Execution (CRITICAL): The command curl -fsSL https://cli.inference.sh | sh is used to install the required CLI. This pattern downloads and immediately executes code from a remote server that is not on the trusted provider list, representing a classic RCE vector.
  • Unverifiable Dependencies & Remote Code Execution (MEDIUM): The skill promotes the use of npx skills add inference-sh/skills@..., which installs additional code from an external, untrusted source into the agent's workspace.
  • Indirect Prompt Injection (LOW): The skill is susceptible to indirect prompt injection as it interpolates user-provided text into shell commands without sanitization.
  • Ingestion points: User prompts for image generation in infsh app run.
  • Boundary markers: None present in the provided shell command examples.
  • Capability inventory: Full access to the infsh tool via Bash.
  • Sanitization: No escaping or validation of user input is described or implemented.
  • COMMAND_EXECUTION (LOW): The skill utilizes Bash(infsh *), granting the agent the ability to execute shell commands. This capability, combined with the lack of input sanitization, creates a risk for command injection if malicious prompts are processed.
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
CRITICAL
Analyzed
Feb 18, 2026, 11:00 PM