Ralph Run
Execute Ralph autonomous iteration loop for a task
Claude CodeGeneric
---
name: /ralph-run
id: ralph-run
category: Ralph
description: Execute Ralph autonomous iteration loop for a task
---
**Usage**
/ralph-run --task
**Flags**
- `--task <id>` (REQUIRED) - Task ID to run (e.g., `parallel-test-execution`)
- `--max-iterations <n>` (optional) - Override max iterations (default: from rulebook.json, usually 10)
- `--pause-on-error` (optional) - Pause loop on first error instead of continuing
- `--dry-run` (optional) - Simulate loop without executing changes
- `--verbose` (optional) - Show iteration details
**What It Does**
Ralph autonomous loop:
1. **Iteration 1**: Analyze problem, design Phase 1
2. **Iteration 2**: Implement Phase 1, test, debug
3. **Iteration 3**: Implement Phase 2, integrate
4. **Iteration N**: Continue until task complete or max iterations reached
Each iteration:
- ✅ Fresh context (avoids AI exhaustion/hallucination drift)
- ✅ Validates compilation (type-check, lint, test)
- ✅ Tracks progress in `.rulebook/ralph/history.json`
- ✅ Can pause/resume across sessions
**Examples**
```bash
# Run with default max iterations (from rulebook.json)
/ralph-run --task parallel-test-execution
# Run with custom max iterations
/ralph-run --task coverage-migration --max-iterations 3 --verbose
# Dry run to see what would happen
/ralph-run --task parallel-test-execution --dry-run
# Pause on error for debugging
/ralph-run --task parallel-test-execution --pause-on-error
Success Criteria Ralph will run until:
- ✅ Task reaches "completed" status, OR
- ⏳ Max iterations reached, OR
- ⚠️ --pause-on-error triggered on failure
Check status with /ralph-status
Related Commands
/ralph-init- Initialize task (generates PRD)/ralph-status- Check current status/ralph-history- View iteration history/ralph-pause- Pause loop gracefully/ralph-resume- Resume from pause
References
Maintain Ralph Run?
Let people know it's listed here — add the badge (live metrics, light/dark aware) or a plain link to your README or docs.
[Ralph Run on getagentictools](https://getagentictools.com/loops/hivellm-run-with-default-max-iterations-from-rulebook-json?ref=badge)