php-code-injection

Fail

Audited by Gen Agent Trust Hub on Jul 6, 2026

Risk Level: CRITICALCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions to execute operating system commands on target systems using PHP functions such as system(), passthru(), exec(), and the shell backtick operator.
  • [REMOTE_CODE_EXECUTION]: Multiple functional reverse shell payloads are included (Bash, native PHP, and Perl), designed to provide interactive shell access to a remote attacker. These payloads triggered a 'reverse_shell_pattern' YARA alert and an antivirus detection (BV:ReverseShell-A [Trj]) for Trojan-like behavior.
  • [DATA_EXFILTRATION]: The skill details methods for exfiltrating sensitive data from targets using network protocols (HTTP and DNS) and file-read functions such as file_get_contents() to access files like /etc/passwd.
  • [PROMPT_INJECTION]: The skill involves processing potentially malicious responses from external targets without sanitization or boundary markers, creating a surface for indirect prompt injection.
  • Ingestion points: External responses from tools like burpsuite and curl (SKILL.md).
  • Boundary markers: Absent.
  • Capability inventory: burpsuite, curl (SKILL.md frontmatter).
  • Sanitization: Absent.
Recommendations
  • CRITICAL: 1 infected file(s) detected - DO NOT USE
  • AI detected serious security threats
Audit Metadata
Risk Level
CRITICAL
Analyzed
Jul 6, 2026, 09:15 AM
Security Audit — agent-trust-hub — php-code-injection