Agent skill
ralph-run
Run RALPH autonomous development loop to implement features from the PRD.
Install this agent skill to your Project
npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/ralph-run-doravidan-supreme-ralph
SKILL.md
Run RALPH
Execute the RALPH autonomous development loop to implement features from the PRD.
Two Operating Modes
1. PRD Loop Mode (Greenfield)
For implementing features from a PRD file:
./scripts/ralph/ralph.sh 20
2. Single-Task Mode (Brownfield)
For quick one-off tasks without a PRD:
./scripts/ralph/ralph.sh --task "Fix the login button styling"
Prerequisites Check
- Verify PRD exists (one of these formats):
cat prd.json 2>/dev/null || cat PRD.md 2>/dev/null || cat tasks.yaml 2>/dev/null
- Check RALPH configuration:
cat .ralph/config.yaml
- Check current status:
node scripts/run-ralph.js --status
Running RALPH
Start the autonomous loop:
# Run up to 20 iterations
./scripts/ralph/ralph.sh 20
# Or use the runner script
node scripts/run-ralph.js 20
With options:
# Skip tests (faster iteration)
./scripts/ralph/ralph.sh --skip-tests 20
# Skip linting
./scripts/ralph/ralph.sh --skip-lint 20
# Custom branch
./scripts/ralph/ralph.sh --branch feature/my-feature 20
# Dry run (no commits)
./scripts/ralph/ralph.sh --dry-run 20
PRD Formats Supported
JSON (prd.json)
{
"userStories": [
{ "id": "US-001", "title": "...", "passes": false }
]
}
Markdown (PRD.md)
## Tasks
- [ ] First task
- [ ] Second task
- [x] Completed task
YAML (tasks.yaml)
tasks:
- title: First task
completed: false
Configuration
Edit .ralph/config.yaml to customize:
- rules: Instructions the AI MUST follow
- boundaries: Files the AI should NOT modify
- commands: Quality gate commands
- settings: Retry logic, auto-commit, etc.
Monitoring
- Watch
.ralph/progress.txtfor iteration logs and learnings - Check PRD file for story completion status
- Review git log for commits:
git log --oneline -20
Commands
| Command | Description |
|---|---|
--status |
Show PRD completion status |
--validate |
Validate PRD schema |
--reset |
Reset progress.txt |
--analyze |
Re-run project analysis |
--task "..." |
Single-task mode (brownfield) |
--skip-tests |
Skip test quality gate |
--skip-lint |
Skip lint quality gate |
--dry-run |
Don't commit changes |
If Stuck
- Check
.ralph/progress.txtfor patterns and learnings - Review
.ralph/config.yamlrules and boundaries - Manually fix blocking issues
- Reset if needed:
node scripts/run-ralph.js --reset - Resume:
./scripts/ralph/ralph.sh 20
Quality Gates
RALPH must pass ALL quality gates before marking a task complete:
- Typecheck:
npm run typecheckornpx tsc --noEmit - Lint:
npm run lint - Tests:
npm test
Configure these in .ralph/config.yaml under commands.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
agent-ops-spec
Manage specification documents in .agent/specs/. Use when user provides requirements, acceptance criteria, or feature descriptions that need to be tracked and validated against implementation.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-testing
Test strategy, execution, and coverage analysis. Use when designing tests, running test suites, or analyzing test results beyond baseline checks.
agent-ops-state
Maintain .agent state files. Use at session start, after meaningful steps, and before concluding: read/update constitution/memory/focus/issues/baseline consistently.
Didn't find tool you were looking for?