ImInDanger

Enforces safety guardrails for the Ralph Wiggum Loop: iteration limits, path restrictions, diff size limits, and validation gates. Use this skill when checking if a loop operation is within safety bounds or when configuring loop safety parameters.

Runs the Ralph Wiggum Loop — an iterative fresh-context agent loop. Use this skill when asked to execute a loop, start iterating on tasks, or run the ralph wiggum pattern. Provides the loop execution script and instructions for invoking it.

Detects convergence issues in Ralph Wiggum Loop iterations: oscillation, stagnation, metric gaming, and context overflow. Use this skill when checking loop health or when the loop seems stuck.

Manages Ralph Wiggum Loop state files (TASKS.md, PROGRESS.md, LOOP_CONFIG.md). Use this skill when asked to initialize loop state, create task files, reset progress, compact progress logs, or manage the loop's persistent disk state.