ralphmode

Warn

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill recommends using CLI flags that explicitly disable or bypass platform security sandboxes and manual approval prompts, such as --dangerously-skip-permissions for Claude Code, --dangerously-bypass-approvals-and-sandbox for Codex CLI, and --yolo for Gemini CLI.
  • [PROMPT_INJECTION]: Contains instructions directed at the agent to operate in 'bypass' or 'YOLO' modes, which are intended to override the default safety and approval constraints of the host platform.
  • [COMMAND_EXECUTION]: Instructs the user to create and execute local shell scripts (ralph-safety-check.sh and ralph-tier1-check.sh) as pre-execution safety hooks. These scripts process tool inputs using python3 logic to enforce custom security boundaries.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 12, 2026, 08:07 AM