ROBOCO CLI

Enterprise-grade scaffolding for full vibe coding — where developers never touch code directly

One command. Your entire repository becomes production-ready for full vibe coding.

npx roboco init

ROBOCO CLI enables full vibe coding at enterprise production level — a development paradigm where developers express intent and AI writes all the code. No manual coding. No code review by humans. Instead, AI agents handle implementation, testing, and validation end-to-end, governed by rigorous process, documentation, and automated quality gates that live in your repository.

This isn't a toy for side projects. ROBOCO sets up the harness, guardrails, and enforcement layer that make full vibe coding safe and reproducible for teams shipping to production.

Why ROBOCO?

The Problem

Full vibe coding demands more discipline than traditional development, not less. Without proper harness engineering:

• AI-generated code with no standardized quality gates ships unchecked to production
• Every developer configures CLAUDE.md, .claude/, MCP servers, and hooks differently
• New team members spend hours figuring out "how do we vibe code here?"
• There's no enforcement — one developer's AI workflow doesn't carry over to another
• Process, documentation, and agent orchestration patterns are tribal knowledge, not code
• Without harness engineering, full vibe coding is a liability, not a superpower

The Solution

ROBOCO turns full vibe coding from a risky experiment into a governed, reproducible engineering practice.

Without ROBOCO	With ROBOCO
AI writes code with no guardrails	Automated quality gates enforce standards on every AI-generated change
Manual CLAUDE.md writing	claude /init + project-specific <roboco> context, auto-generated
Copy-paste .claude/ settings	Stack-detected hooks (TS→Prettier, Py→Black, Go→gofmt)
"Ask Sarah how she set up MCP"	roboco install — one command, identical environment
No process, no docs	5-stage vibe coding templates (Intent → Requirements → Research → Plan → Implement)
Works on my machine	CI/CD enforces vibe coding rules via pre-commit hooks + GitHub Actions
No visibility into setup quality	roboco audit scores your maturity across 5 categories (100 points)
Full vibe coding feels unsafe	Harness engineering makes it production-safe

Key Benefits

For Individual Developers

• Zero-config start — npx roboco init --auto analyzes your repo and sets up everything in seconds
• Best-practice defaults — OMC agents, MCP servers, Claude Code Hooks, all pre-configured for your stack
• Structured process — 5-stage templates prevent "just start coding" anti-patterns

For Teams

• One setup, every developer — roboco install reproduces the exact vibe coding environment from .roboco/config.json
• Enforced consistency — Pre-commit hooks and CI pipelines validate that rules are followed, even by non-vibe-coders
• Drift detection — roboco sync --check catches when config diverges from the repo state
• Onboarding in minutes — New member? git clone && npx roboco install. Done.

For Tech Leads

• Maturity scoring — roboco audit quantifies your vibe coding setup across Claude Code env, process docs, quality gates, tool integration, and team consistency
• Actionable improvements — Every audit finding comes with a specific suggestion
• Standards as code — Vibe coding rules live in the repo, not in a wiki nobody reads

Architecture: Multi-Layer Verification Harness

ROBOCO CLI sets up a 7-layer verification harness that governs the entire AI development lifecycle:

Layer	Gate	What it enforces
1	Developer Intent	5-stage vibe coding process (Intent → Requirements → Research → Plan → Implement)
2	AI Code Generation	CLAUDE.md context, stack-specific hooks, OMC agents, Harness teams
3a	Pre-commit	lint-staged → ESLint + Prettier auto-format
3b	Pre-push	TypeCheck → Tests → AI Code Review (Claude Agent SDK)
4	CI Pipeline	Lint, TypeCheck, Unit, Integration, E2E, Build + Smoke test
5	Release Gate	Tag → Validate → npm publish → GitHub Release
6	Repo Hygiene	Issue auto-label, stale management, drift detection
7	Maturity Audit	100-point scoring across 5 categories

• Maturity scoring — roboco audit quantifies your vibe coding setup across Claude Code env, process docs, quality gates, tool integration, and team consistency
• Actionable improvements — Every audit finding comes with a specific suggestion
• Standards as code — Vibe coding rules live in the repo, not in a wiki nobody reads

Features

• AI-Powered Interview — Analyzes your repo (stack, structure, existing config) then asks targeted questions via Claude Code SDK to customize the setup
• Smart Stack Detection — Automatically detects your tech stack (TypeScript, Python, Go, Rust, Java, etc.) and applies optimized settings
• Claude Code Environment — Delegates to claude /init for CLAUDE.md, generates .claude/ settings, Claude Code Hooks, and MCP server configurations
• Team Consistency — Every developer who clones the repo gets the identical vibe coding environment via roboco install
• CI/CD Enforcement — Pre-commit hooks and GitHub Actions workflows enforce vibe coding rules even for non-vibe-coding members
• 5-Stage Process Templates — Intent → Requirements → Research → Plan → Implement, with documents at each stage
• Maturity Audit — Score your vibe coding setup across 5 categories with actionable improvement suggestions
• Drift Detection — roboco sync catches configuration drift; --check flag for CI integration
• Modular Tool Integration — Add tools incrementally with roboco add (OMC, Exa.ai, Context7, GitHub MCP, Harness, OpenSpec)

Quick Start

Initialize a new project

npx roboco init

This will:

1. Analyze your repository structure and tech stack
2. Conduct an AI interview to understand your preferences
3. Generate CLAUDE.md, .claude/ configuration, and hooks
4. Install required tools (oh-my-claudecode, MCP servers)
5. Set up optional process documents and CI/CD pipelines

Auto mode (non-interactive)

npx roboco init --auto

AI suggests the optimal setup based on analysis. You just approve or reject.

Dry run

npx roboco init --dryrun

Analyze and show recommendations without making any changes.

Commands

Command	Description
roboco init [path]	Initialize vibe coding environment with AI interview
roboco install [path]	Apply existing config for team members (no interview)
roboco update [path]	Update existing setup with new analysis
roboco status [path]	Report current vibe coding setup status
roboco doctor	Diagnose ROBOCO CLI health (version, SDK, dependencies)
roboco config	View and modify global ROBOCO configuration
roboco add <integration>	Add a tool post-init (openspec, exa, github, context7, harness)
roboco sync [--check]	Detect configuration drift (CI-friendly with --check)
roboco validate [--fix]	End-to-end setup validation
roboco audit [--format]	Vibe coding maturity scoring (100 points, 5 categories)

What Gets Generated

Required (always created)

Artifact	Description
CLAUDE.md	AI context document tailored to your repo
.claude/settings.json	Claude Code settings with stack-specific hooks
.claude/ directory	Commands, skills, and configuration
oh-my-claudecode	Multi-agent orchestration (installed as dependency)

Optional (selected during interview)

Artifact	Description
OpenSpec	5-stage vibe coding process documentation framework
Exa.ai MCP	Code/technical documentation web search
Perplexity MCP	General web search and Q&A (fallback)
GitHub MCP	Issue/PR management and code search
Context7 MCP	Up-to-date library/framework documentation
Harness	Domain-specific agent team design (6 architecture patterns)
CI/CD pipeline	Pre-commit hooks + GitHub Actions workflows
Process templates	5-stage vibe coding document templates

Tech Stack

• Language: TypeScript (ESM)
• Runtime: Node.js >= 24
• CLI Framework: Commander.js
• AI Integration: Claude Code SDK
• Distribution: npm (npx roboco init)

Requirements

• Node.js 24 or later
• Claude subscription or API key (required for Claude Code SDK)

Development

# Install dependencies
npm install

# Build
npm run build

# Run tests
npm run test

# Development mode (watch)
npm run dev

Roadmap

v0.1 — MVP

• 6 core commands: init, install, update, status, doctor, config
• AI interview (Claude Agent SDK) with auto/interactive/dryrun modes
• Stack-agnostic configuration with auto-detection (7 languages, 9 frameworks)
• OMC + MCP servers + Harness + OpenSpec integration
• npm distribution (npx roboco init)

v0.2 — Extensibility

• add, sync, validate commands
• Pre-commit hook generation (stack-specific via husky)
• CLAUDE.md delegates to claude /init + <roboco> context

v0.3 — Audit & Automation

• audit command (100-point vibe coding maturity scoring)
• Release automation (tag → validate → npm publish → GitHub Release)
• Issue auto-labeling and stale management

v0.4 — Current

• Existing .claude/ and ~/.claude/ settings analysis and deep merge
• AI code review on pre-push (Claude Agent SDK)
• DESIGN.md, CONTRIBUTING.md, CHANGELOG.md, LICENSE
• CI paths-ignore for non-executable changes
• 62 tests (unit + integration + E2E)

Future

• eject, share commands
• Stack-specific configuration presets
• Community preset/plugin ecosystem

Contributing

This project is 100% vibe-coded — every line of code, every test, every document has been written by AI agents. We welcome contributions from both humans and AI.

• Feature Requests — Open an issue describing what you want. AI-generated FRs with detailed specs are welcome.
• Pull Requests — AI-authored PRs are first-class citizens here. Whether you vibe-coded it with Claude Code, Cursor, Copilot, or any other AI tool, send it.
• Bug Reports — Describe the issue. AI-generated reproduction steps are appreciated.

See CONTRIBUTING.md for development setup, commit conventions, and the anti-reinvention rule.

License

MIT

Links

• Product Requirements Document
• Research Report
• Quality Gates
• Changelog
• Contributing