ai-image-generation

Fail

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The installation guide in SKILL.md uses the command curl -fsSL https://cli.inference.sh | sh. This pattern is high-risk because it executes a remote script directly in the shell without any verification or oversight, allowing the source to run arbitrary code on the host machine.
  • [EXTERNAL_DOWNLOADS]: The 'Related Skills' section suggests adding multiple external skills via npx skills add inference-sh/skills@.... This fetches and executes code from unverified third-party repositories on the npm registry, which could lead to supply chain attacks.
  • [COMMAND_EXECUTION]: The skill explicitly allows the use of the Bash(infsh *) tool. Since the infsh binary is installed via an unverified remote script, the agent is granted the capability to execute commands from a potentially compromised or malicious source.
  • [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8). Ingestion points: The --input flag in infsh app run (SKILL.md) accepts JSON objects containing user-controlled prompt strings. Boundary markers: There are no delimiters or specific instructions provided to the agent to treat the input as data rather than potential commands. Capability inventory: The skill allows command execution via Bash(infsh *). Sanitization: The instructions do not include any steps to sanitize or escape the user input before it is passed to the CLI.
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
Mar 12, 2026, 07:36 AM