Root terminal for any LLM.
RootClaude is an open-source coding-agent CLI with 19 premium skills, 8 free models, and deep reasoning capabilities. Built for software engineers who want powerful AI assistance without the cost.
░▒▓███████▓▒░ ░▒▓██████▓▒░ ░▒▓██████▓▒░▒▓████████▓▒░▒▓███████▓▒░░▒▓█▓▒░░▒▓█▓▒░░▒▓██████▓▒░ ░▒▓███████▓▒░
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░ ░▒▓█▓▒░
░▒▓███████▓▒░░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓███████▓▒░░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒▒▓███▓▒░░▒▓██████▓▒░
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░ ░▒▓█▓▒░
░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░ ░▒▓█▓▒░ ░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░▒▓█▓▒░░▒▓█▓▒░ ░▒▓█▓▒░
░▒▓█▓▒░░▒▓█▓▒░░▒▓██████▓▒░ ░▒▓██████▓▒░ ░▒▓█▓▒░ ░▒▓███████▓▒░ ░▒▓██████▓▒░ ░▒▓██████▓▒░░▒▓███████▓▒░
- 19 Premium Skills — Code quality, design, security, performance, reasoning
- 8 Free Models — MiMo, Qwen, DeepSeek, Nemotron, GPT-OSS, Gemma, North
- Zero Cost — All models are free tier, no API keys required
- Smart Routing — Simple tasks → fast model, complex tasks → powerful model
- Deep Reasoning — Multi-step thinking, chain-of-thought, self-verification
- AutoCLI WebFetch — Direct HTTP fetch without browser dependency
- Persistent Memory — Cross-session context and learnings
git clone https://github.com/RootBugs/Rootclaude.git
cd Rootclaude
npm install
npm run build
node bin/rootclauderootclaudeRootClaude comes with OpenCode Zen as the default provider — no API key setup required.
| Setting | Value |
|---|---|
| Provider | OpenCode Zen |
| Model | DeepSeek V4 Flash Free |
| Endpoint | https://opencode.ai/zen/v1 |
| API Key | Not required (free tier) |
Simply run rootclaude and start coding. No configuration needed.
| Category | Skills |
|---|---|
| Code Quality | elite-coder, ultra-prompt, code-quality, tdd-enforcer |
| Design | premium-design, ui-design-quality, responsive-design |
| Security | security-first |
| Performance | performance-optimized, agent-cache, context-optimizer |
| Memory | smart-memory |
| Search | multi-search |
| Reasoning | deep-reasoning, system-prompt-master |
| Development | fullstack-master, react-nextjs-master, project-builder, debug-fix |
| Utilities | ultra-fast, css-validator, verify-claims, render-check, memory-manager, agent-orchestrator |
| Agent Type | Model | Provider |
|---|---|---|
| Explore | MiMo V2.5 Free | OpenCode Zen |
| Plan | Qwen3 Coder | OpenRouter |
| General | MiMo V2.5 Free | OpenCode Zen |
| Verification | Nemotron 3 Super | OpenRouter |
| Default | DeepSeek V4 Flash Free | OpenCode Zen |
| Model | Provider | Type | Cost |
|---|---|---|---|
| DeepSeek V4 Flash Free | OpenCode Zen | Main | Free |
| MiMo V2.5 Free | OpenCode Zen | Sub-agent | Free |
| Qwen3 Coder | OpenRouter | Sub-agent | Free |
| Nemotron 3 Super | OpenRouter | Sub-agent | Free |
| Llama 3.3 70B | OpenRouter | Sub-agent | Free |
| GPT-OSS 120B | OpenRouter | Sub-agent | Free |
| Gemma 4 31B | OpenRouter | Sub-agent | Free |
| North Mini Code | OpenRouter | Sub-agent | Free |
RootClaude supports 200+ models across multiple providers:
| Provider | Models | API Key Required |
|---|---|---|
| OpenCode Zen | DeepSeek, MiMo, Qwen, GLM, MiniMax, Kimi | No (free tier) |
| Anthropic | Claude Opus, Sonnet, Haiku | Yes |
| OpenAI | GPT-5, GPT-4o, Codex | Yes |
| Google Gemini | Gemini 2.0, Gemini Pro | Yes |
| Ollama | Llama, Mistral, Qwen (local) | No |
| LM Studio | Any local model | No |
| GitHub Copilot | GPT-4o, Claude | Yes |
| AWS Bedrock | Claude, Llama, Mistral | Yes |
| Google Vertex | Gemini, Claude | Yes |
| DeepSeek | DeepSeek V4, V3 | Yes |
Switch providers with /provider or set environment variables:
# OpenCode Zen (default - free)
export OPENAI_BASE_URL=https://opencode.ai/zen/v1
export OPENAI_API_KEY=public
export OPENAI_MODEL=deepseek-v4-flash-free
# OpenAI
export CLAUDE_CODE_USE_OPENAI=1
export OPENAI_API_KEY=sk-your-key
export OPENAI_MODEL=gpt-4o
# Anthropic
export ANTHROPIC_API_KEY=sk-ant-your-keyConfigure via settings.json in ~/.openclaude/:
{
"model": "deepseek-v4-flash-free",
"permissions": {
"allow": ["Bash(npm:*)", "Bash(git:*)", "Read", "Edit", "Write"],
"deny": ["Bash(rm -rf /:*)"]
},
"hooks": {},
"env": {}
}RootClaude
├── Main Model (DeepSeek V4 Flash Free via OpenCode Zen)
├── 8 Agent Models (OpenRouter)
├── 19 Skills (reasoning-enhanced)
├── Agent Routing (smart model selection)
├── Smart Routing (simple → fast, complex → powerful)
├── AutoCLI WebFetch (direct HTTP)
├── Persistent Memory (cross-session)
└── GitHub: RootBugs/Rootclaude
| Command | Description |
|---|---|
/help |
Show available commands |
/status |
Show current status |
/model |
Switch model |
/provider |
Switch provider |
/skills |
List available skills |
/compact |
Compact conversation |
/clear |
Clear conversation |
/theme |
Change theme |
/hooks |
Manage hooks |
/mcp |
MCP server management |
MIT — See LICENSE for details.
Built with ❤️ by RootBugs