interactive-json-form

Fail

Audited by Gen Agent Trust Hub on Aug 11, 2026

Risk Level: HIGHCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions to execute multi-line shell scripts on a remote server ('bigblack') via SSH. This mechanism allows for arbitrary command execution on the target host.
  • [PRIVILEGE_ESCALATION]: Deployment steps require using sudo on a remote system to perform administrative tasks such as directory creation and ownership changes. The skill explicitly notes the existence of passwordless sudo privileges for the executing user.
  • [COMMAND_EXECUTION]: The skill uses shell tools (scp, ssh, curl) to manage file lifecycle and connectivity across remote systems ('bigblack' and 'aws-dev').
  • [DATA_EXFILTRATION]: The skill automates the transfer of generated files to a publicly accessible web domain ('eon.25u.com'). While the skill attempts to obfuscate the location using randomized paths and 'noindex' headers, it effectively moves data to the public internet.
  • [DYNAMIC_EXECUTION]: Deployment paths are constructed dynamically at runtime using openssl rand -hex 16 and then interpolated into shell commands, increasing the complexity of auditing the final execution path.
  • [INDIRECT_PROMPT_INJECTION]: The HTML template includes a client-side JSON import feature (btnImport) that parses external files into the application state, creating an attack surface for processing untrusted data.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Aug 11, 2026, 05:05 PM
Security Audit — agent-trust-hub — interactive-json-form