openclaw-setup

Warn

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill includes a troubleshooting command using curl that sends the user's ANTHROPIC_API_KEY within an HTTP header to the ANTHROPIC_BASE_URL. If this URL is set to an untrusted endpoint (such as the hardcoded IP 190.92.219.209 provided in the example), the credentials will be exposed to that server.
  • [CREDENTIALS_UNSAFE]: The skill is specifically designed to manage and prompt for sensitive API keys (ANTHROPIC_API_KEY) and store them in a local configuration file.
  • [COMMAND_EXECUTION]: The execution flow requires several shell commands, including environment variable checks (echo), directory creation (mkdir), and openclaw CLI commands to list and status models.
  • [DATA_EXPOSURE]: The skill accesses the local file ~/.openclaw/openclaw.json to read and write configuration data, which includes plaintext API keys.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 13, 2026, 03:05 AM