mylukin/ralph-dev/dev-orchestrator
Claude Code and agent skill by mylukin/ralph-dev
Trust tier 3
- community
Source check pending
Checking that the SKILL.md file is still in the repository.
Autonomous end-to-end development from requirement to delivery. Use when user wants complete automation, "build X for me", or full feature implementation without manual steps.
Install
npx vskill@latest install mylukin/ralph-dev/dev-orchestratorSKILL.md
Ralph-dev Orchestrator
Goal
Transform a user requirement into delivered, tested, production-ready code with ZERO manual intervention after initial clarification.
Workflow Phases
State Management
All state persists in .ralph-dev/:
state.json- Current phase, progress (managed by CLI)prd.md- Product requirements documenttasks/- Task files with index.json
Execution
Initialize
Main Loop (Context-Compression Resilient)
Phase Invocation
Use Task tool to invoke each phase skill:
Phase Summary
| Phase | Skill | Interactive | Key Output | |
|---|---|---|---|---|
| 1. Clarify | phase-1-clarify | Yes | .ralph-dev/prd.md | |
| 2. Breakdown | phase-2-breakdown | Yes (approval) | .ralph-dev/tasks/ | |
| 3. Implement | phase-3-implement | No | Code + tests | |
| 4. Heal | phase-4-heal | No | (invoked by Phase 3) | |
| 5. Deliver | phase-5-deliver | Optional | Commit + PR |
Mode Commands
| Command | Action | |
|---|---|---|
/ralph-dev {requirement} | Start new session | |
/ralph-dev resume | Continue from saved state | |
/ralph-dev status | Show current progress | |
/ralph-dev cancel | Archive and clear session |
State Transitions
- Each phase updates state via
ralph-dev state update --phase {next} - Phase 4 (heal) is invoked on-demand, not a state transition
- Phases can only move forward (except implement ⇄ heal loop)
Safety Limits
| Limit | Value | Purpose | |
|---|---|---|---|
| Orchestrator timeout | 12 hours | Prevent infinite loops | |
| Per-phase timeout | Varies | Defined in each phase skill | |
| Heal attempts | 3 per task | Circuit breaker |
Constraints
- NEVER skip phases (must complete in order)
- NEVER rely on memory variables (always query CLI)
- ALWAYS get user approval before implement phase
- ALWAYS save state after each phase completion
- ALWAYS show progress updates during long operations
Error Handling
| Error | Action | |
|---|---|---|
| Phase fails | Log error, show diagnostics, don't auto-retry | |
| User cancels | Save state, show resume command | |
| Interrupted | State persists, resume on next session | |
| Unknown phase | Report error, suggest manual state reset |
Resume Behavior
When resuming from each phase:
| Phase | Resume Action | |
|---|---|---|
| clarify | Continue with remaining questions | |
| breakdown | Show plan again for approval | |
| implement | Get next task via ralph-dev tasks next | |
| deliver | Re-run delivery checks |
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryCheckingskills/dev-orchestrator/SKILL.md
Popularity
- Repository stars
- 34
- Repository forks
- 3
- 7-day trend score
- 0
Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/mylukin/ralph-dev/dev-orchestrator)