overcome-false-limitations

Warn

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: The skill explicitly directs the agent to locate and read 'credentials already present' and browser 'cookies' as part of its primary evidence collection strategy to resolve authentication failures.- [DATA_EXFILTRATION]: Instructions in the Evidence Order section require the agent to inspect sensitive local state, including environment variables (env), browser profiles, and configuration files, to identify failing layers during troubleshooting.- [PROMPT_INJECTION]: The 'Core Rule' and 'Anti-Gate Rule' instruct the agent to disregard and re-test previous system or tool conclusions like 'blocked', 'not possible', or 'requires auth', effectively encouraging the agent to bypass reported security or access constraints.- [COMMAND_EXECUTION]: The skill recommends executing commands through interactive login shells (zsh -lic) to recover from PATH failures, which may result in commands being executed with expanded environment privileges or unexpected shell configurations.- [PROMPT_INJECTION]: The skill operates as a high-risk surface for indirect prompt injection because it ingests untrusted data from command logs, tool traces, and browser outputs without utilizing boundary markers or sanitization to prevent the agent from following embedded instructions.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Aug 4, 2026, 12:03 PM