ralph-wiggum-looplisted

# Ralph Wiggum Loop The autonomous task completion loop - the heart of Gold tier autonomy. ## What It Does Continuously monitors `/Needs_Action` folder and autonomously: 1. Reads pending tasks 2. Analyzes task requirements 3. Creates execution plans 4. Executes safe actions automatically 5. Requests approval for sensitive actions 6. Logs all activities 7. Self-heals on errors ## Features - **Autonomous Operation:** Runs every 10 minutes - **Smart Decision Making:** Knows what needs approval vs auto-execute - **Self-Healing:** Recovers from errors automatically - **Audit Trail:** Complete logging of all actions - **Safety First:** Sensitive actions always require approval ## Usage ### Start the loop (recommended) Use Claude Code's `/loop` command to run continuously: ```bash /loop 10m python .claude/skills/ralph-wiggum-loop/scripts/ralph_loop.py ``` This will run the loop every 10 minutes automatically. Claude Code manages scheduling and restarts. ### Run once (test mode) ```bash python .claude/skills/ralph-wiggum-loop/scripts/ralph_loop.py ``` ### Stop the loop Stop the `/loop` command in Claude Code interface. ## How It Works ``` Every 10 minutes: 1. Check /Needs_Action for tasks 2. For each task: - Read and analyze - Determine if safe to auto-execute - If safe: Execute and move to /Done - If sensitive: Create approval request in /Pending_Approval - If error: Retry with backof