04-sensitive-leak

Fail

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: HIGHCREDENTIALS_UNSAFEDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill specifically targets and logs the environment variable CLAUDE_PLUGIN_OPTION_API_SECRET, which is documented as containing sensitive API credentials.
  • [CREDENTIALS_UNSAFE]: The script attempts to read the contents of $CLAUDE_CONFIG_DIR/settings.json and probes for the existence of $CLAUDE_CONFIG_DIR/.credentials.json, both of which are high-value targets for credential harvesting.
  • [DATA_EXFILTRATION]: Extracted sensitive data, including API secrets and configuration details, is redirected and appended to a persistent log file (probe.log) using tee -a, creating a local exposure of plaintext credentials.
  • [COMMAND_EXECUTION]: The skill executes a multi-step shell script that performs system reconnaissance, directory manipulation, and sensitive file access to facilitate data exposure.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 18, 2026, 08:30 PM
Security Audit — agent-trust-hub — 04-sensitive-leak