# v2 git bundle
a2a911d1f83aa0a3fe844561edc141eb06d871ea refs/heads/bundle
a2a911d1f83aa0a3fe844561edc141eb06d871ea HEAD

PACK      ˜xø ÿtree b3e94acb822c71780b25957f2126970e528edc07
parent 49be6b9fc41b1fe992ce2654b9220c0fb5fca20e
author Hanzo Dev <dev@hanzo.ai> 1772248388 -0800
committer Hanzo Dev <dev@hanzo.ai> 1772248388 -0800

chore: add LLM.md knowledge base, update .gitignore
fïJË“xã ÿtree 4bfbf790392a98cf288f5b4bc21a44534ae2a5b4
parent 364e7ed29ad4ae33ec3d96e6afc765b0ca6f607f
author z <z@zeekay.ai> 1772233474 -0800
committer z <z@zeekay.ai> 1772233474 -0800

fix: remove upstream model references from READMEiE–xö 	ÿtree 4bfbf790392a98cf288f5b4bc21a44534ae2a5b4
parent 6c55125cfc7e202ae34aa55bf95f63a1b9506830
author Hanzo Dev <dev@hanzo.ai> 1772232581 -0800
committer Hanzo Dev <dev@hanzo.ai> 1772232581 -0800

fix: remove upstream model references from README
ÊJbœxŒ sÿtree c5b0d462e5fdc89e2a2177b6053ea8da7ee96881
author z <z@zeekay.ai> 1772164244 -0800
committer z <z@zeekay.ai> 1772164244 -0800

Add README]W'º­xm ’ÿ100644 .gitignore NA-4xÐƒÿ[í%ÚÈf€100644 LLM.md pRÝyÝD›e4©-tåJºÆ:Í100644 README.md æ›_(LÞÐ3åWlb?5ÐoŸõax)¼xL ³ÿCLAUDE.md
AGENTS.md
GEMINI.md
QWEN.md
*.safetensors
*.bin
*.gguf
*.pt
*.pth
OI2¶2x&Ùü# zen-code â€” AI Knowledge Base

**Project**: zen-code
**Organization**: zenlm
**Repository**: https://github.com/zenlm/zen-code
**HuggingFace**: https://huggingface.co/zenlm/zen-code
**Last Updated**: 2026-02-27

## Overview

zen-code is part of the Zen AI model family by Hanzo AI / Zen LM.

## Rules for AI Assistants

1. **ALWAYS** update LLM.md with significant discoveries
2. **NEVER** commit model weights (*.safetensors, *.bin, *.gguf, *.pt)
3. **NEVER** commit symlinked files (CLAUDE.md, AGENTS.md, GEMINI.md, QWEN.md)
4. **NEVER** create random summary files â€” update THIS file only
5. Zen models are based on **Qwen3** architecture

## Context

This file (`LLM.md`) is symlinked as CLAUDE.md, AGENTS.md, GEMINI.md, QWEN.md.

---

*Part of the Zen AI family â€” Clarity Through Intelligence*
Jßj²«x²
Mõ# Zen Code â€” AI Coding Agent for Your Terminal

An open-source AI coding agent that lives in your terminal. Powered by the Zen Coder model family from [Zen LM](https://zenlm.org).

[![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
[![Node.js Version](https://img.shields.io/badge/node-%3E%3D20.0.0-brightgreen.svg)](https://nodejs.org/)

## Why Zen Code?

- **Codebase understanding**: Ask questions about large, unfamiliar repos
- **Multi-file edits**: Refactor across files with a single prompt
- **Test execution**: Run tests and iterate on failures automatically
- **Git integration**: Commit, diff, and review changes in-flow
- **MCP tools**: 260+ tools via Model Context Protocol
- **Intelligent routing**: Automatically selects from Zen Coder 4Bâ€“480B based on task complexity

## Installation

```bash
npm install -g @zen/code
```

Or with Homebrew:

```bash
brew install zen-code
```

## Quick Start

```bash
# Launch interactive session
zen-code

# Run a single task (headless)
zen-code "add tests for the auth module"

# Reference specific files
zen-code "refactor @src/api/handler.ts to use async/await"
```

## Usage

### Interactive mode

```bash
cd your-project/
zen-code
```

Type prompts directly. Use `@filename` to reference files in context.

### Headless mode (scripts, CI)

```bash
zen-code -p "fix the failing tests in src/"
```

### IDE integration

Zen Code integrates with VS Code, Zed, and JetBrains IDEs. See [zenlm.org](https://zenlm.org) for setup guides.

## Configuration

Create `~/.zen/settings.json`:

```json
{
  "modelProviders": {
    "openai": [
      {
        "id": "zenlm/zen-coder",
        "name": "Zen Coder",
        "baseUrl": "http://localhost:8000/v1",
        "envKey": "ZEN_API_KEY"
      }
    ]
  },
  "model": {
    "name": "zenlm/zen-coder"
  }
}
```

## Models

Zen Code uses the Zen Coder family with intelligent routing:

| Model | Parameters | Context | Best For |
|-------|-----------|---------|----------|
| Zen Coder 4B | 4B | 32K | Fast edits, edge deployment |
| Zen Coder Flash | 31B MoE | 131K | Balanced performance |
| Zen Coder 480B | 480B MoE | 128K | Complex refactors, frontier tasks |

Deploy with vLLM:

```bash
vllm serve zenlm/zen-coder --port 8000
```

## Session Commands

- `/help` â€” list commands
- `/clear` â€” clear conversation history
- `/compress` â€” compress history to save tokens
- `/stats` â€” show session info
- `/exit` â€” quit

## Links

- **Website**: https://zenlm.org
- **Models**: https://huggingface.co/zenlm
- **GitHub**: https://github.com/zenlm
- **Hanzo MCP**: https://github.com/hanzoai/mcp

## License

Apache 2.0

---

**Zen AI**: Clarity Through Intelligence
¿Ø€p¥x% Úÿ100644 README.md æ›_(LÞÐ3åWlb?5ÐoŸõ×N¥x% Úÿ100644 README.md ¯`>ÌPö°ðþi»Ù @ÀÒ°K5ì—´?xôü# Zen Code 4B

General-purpose code generation model. Part of the Zen Eco family.

[![License](https://img.shields.io/badge/License-Apache_2.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)

## Overview

Zen Code 4B is a compact code generation model optimized for fast inference. Supports multiple programming languages with strong performance on code completion, generation, and explanation tasks.

| Property | Value |
|----------|-------|
| Parameters | 4B |
| Context | 32K |
| License | Apache 2.0 |

## Usage

```python
from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("zenlm/zen-code")
tokenizer = AutoTokenizer.from_pretrained("zenlm/zen-code")
```

## Related

- [zen-eco](https://huggingface.co/zenlm/zen-eco) â€” Base 4B model
- [zen-coder](https://huggingface.co/zenlm/zen-coder) â€” 24B code model
- [Zen LM](https://github.com/zenlm) â€” Full model family

Apache 2.0 Â· [Zen LM](https://zenlm.org) Â· [Hanzo AI](https://hanzo.ai)
x`^ÚÂb
ÿ	<¾
—|;$K{³¼l