Skip to content
inferred from GitHub

EVERYTHING CLAUDE FREE VERSION

Code Cerebral Valley

Core trick

Wires Claude Code agent tasks to Gemma 4 via Google AI Studio and small tasks to a local Qwen model through Ollama.

Evidence

README matched ""Cerebral Valley" "Opus 4.6"": docs/zh-CN/README.md

Tags

dev tools
Krishnendu409/EVERYTHING-CLAUDE-FREE-VERSION★ 10 forksJavaScriptMITCreated Apr 10, 2026 · last push Oct 16, 2026

README excerpt

# ECC — Free Version

> The full **Everything Claude Code** project ([affaan-m/everything-claude-code](https://github.com/affaan-m/everything-claude-code)) — all agents, skills, hooks, rules, and commands — pre-wired for **free models** instead of paid Claude APIs.

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](LICENSE)

---

## Models

| Role | Model | Provider | Cost |
|------|-------|----------|------|
| Primary / large tasks | **Gemma 4 27B** (`google/gemma-4-27b-it`) | Google AI Studio | Free (rate-limited) |
| Fast / small tasks | **Qwen 2.5 Coder 7B** (`ollama/qwen2.5-coder:7b`) | Ollama (local) | Free (local GPU/CPU) |

---

## Quick Setup

### 1. Get a Google AI Studio key
Go to <https://aistudio.google.com> → Get API key → copy it.

### 2. Export the key
```bash
export GOOGLE_AI_STUDIO_KEY=your_key_here
```
Or copy `.env.example` to `.env` and fill it in (requires your shell to source it).

### 3. Install Ollama + pull the local model
```bash
bash scripts/setup-ollama.sh
```
This installs Ollama if needed and pulls `qwen2.5-coder:7b` (~4.7 GB).

### 4. Install OpenCode
```bash
npm i -g opencode
```

### 5. Run
```bash
opencode
```

---

## Switching Models

Edit `.opencode/opencode.json` — change the top-level `"model"` field:

```jsonc
// Use Gemma (default — needs GOOGLE_AI_STUDIO_KEY):
"model": "google/gemma-4-27b-it"

// Switch to fully-local Qwen when quota is exceeded:
"model": "ollama/qwen2.5-coder:7b"
```

---

## Check Your Quota
``