GitHub - robotlearning123/nano-opencode

A minimal yet powerful AI coding assistant for the terminal.

Quick Start

# Install (bun recommended for speed)
git clone https://github.com/robotlearning123/nano-opencode.git
cd nano-opencode && bun install

# Set API key (choose one)
export ANTHROPIC_API_KEY=your_key
export OPENAI_API_KEY=your_key
export GEMINI_API_KEY=your_key

# Run
bun dev                    # Development (fastest, no build)
bun run build && bun start # Production

Features

Feature	Description
Multi-Provider	Anthropic, OpenAI, Gemini
16 Tools	File ops, search, bash, diff, patch, LSP, and more
10 Agents	Specialized agents for different tasks
Hooks	Lifecycle hooks for extensibility
Skills	Template-based prompt system
MCP	Model Context Protocol support
Sessions	SQLite-persisted conversation history
Security	Path traversal protection, dangerous command detection

Usage

Interactive Mode

npm start        # or: node dist/app.js

Single Prompt

npm start -- --prompt "explain this codebase"

Commands

Command	Description
`/help`	Show help
`/agent [name]`	List or switch agents
`/tools`	List available tools
`/sessions`	List recent sessions
`/new`	Start new session
`/mcp`	List MCP servers
`/skill [name]`	List or show skills
`/connect [provider]`	Configure provider auth
`/quit`	Exit

Tools

Tool	Description
`read_file`	Read file with line numbers
`write_file`	Create or overwrite file
`edit_file`	Find and replace in file
`bash`	Execute shell command
`glob`	Find files by pattern
`grep`	Search file contents
`list_dir`	List directory
`diff`	Compare files or content
`patch`	Atomic multi-file edits
`todo`	Manage task list
`webfetch`	Fetch URL content
`skill`	Execute a skill
`session`	Session management
`lsp_*`	LSP code intelligence

Agents

Agent	Purpose
`sisyphus`	Full-capability orchestrator (default)
`oracle`	Read-only advisor
`librarian`	Documentation explorer
`explore`	Codebase search
`junior`	Simple task executor
`prometheus`	Strategic planner
`metis`	Pre-planning consultant
`momus`	Plan reviewer
`frontend`	UI development
`multimodal`	Image/PDF analysis

Configuration

Create ~/.config/nano-opencode/config.json:

{
  "provider": "anthropic",
  "model": "claude-sonnet-4-20250514",
  "maxTokens": 8192,
  "temperature": 0.7
}

Or use environment variables:

export ANTHROPIC_API_KEY=sk-...
export OPENAI_API_KEY=sk-...
export GEMINI_API_KEY=AI...

Architecture

src/
├── app.ts           # Entry point
├── cli.ts           # Interactive CLI
├── config.ts        # Configuration
├── store.ts         # SQLite persistence
├── utils.ts         # Security utilities
├── agents/          # 10 built-in agents
├── hooks/           # Lifecycle hooks
├── skills/          # Skill templates
├── providers/       # Anthropic, OpenAI, Gemini
├── tools/           # 16 tools
├── mcp/             # Model Context Protocol
└── ui/              # Terminal formatting

Development

bun dev          # Run directly (no build needed)
bun run test     # Run tests (74 passing)
bun run build    # Build TypeScript
bun run typecheck # Type check only

Stats

Metric	Value
Lines	~7,000 TypeScript
Files	51 source files
Tests	74 passing
Startup	<70ms (bun)
Install	~5s (bun)

License

MIT

Simple tools, clear prompts, powerful results.