using-superpowers

Fail

Audited by Gen Agent Trust Hub on Jun 13, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The SKILL.md file employs aggressive behavioral manipulation to override the agent's default reasoning and system constraints. It uses directives like "ABSOLUTELY MUST," "not negotiable," and "not optional" to force the agent to invoke the Skill tool even for simple clarifying questions. This mirrors techniques used in prompt injection to bypass safety filters and default behavioral guidelines.
  • [COMMAND_EXECUTION]: The references/codex-tools.md file provides instructions to modify the local system's configuration (~/.codex/config.toml). This involves enabling features like multi_agent, which changes how the environment operates. Instructions that lead to the modification of system or application configuration files can be used to establish persistence or degrade the host's security profile.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Jun 13, 2026, 05:19 PM