Jada Code

AI coding agent for Garza OS — powered by Goose

Jada Code is a custom distribution of Goose — the open-source AI agent by AAIF (Agentic AI Foundation). It's preconfigured for the Garza OS ecosystem with Kimi K2.5 via OpenRouter, MCP extensions for Nextcloud, and server-first deployment.

What's different from upstream Goose

• Branded as Jada Code — system prompt, Dockerfile, metadata
• Preconfigured for Garza OS — OpenRouter provider, Kimi K2.5 model
• Server-first deployment — Docker Compose for headless VPS operation
• MCP extensions — Nextcloud Passwords, Nextcloud files, and more

Quick Start (Server)

# Clone
git clone https://github.com/itsablabla/jada-code.git
cd jada-code

# Run the server
docker compose -f docker-compose.server.yml up -d

# Server available at http://localhost:3000

Quick Start (CLI)

# Build from source
cargo build --release --package goose-cli

# Run
GOOSE_PROVIDER=openrouter \
OPENROUTER_API_KEY=your-key \
./target/release/goose session

Configuration

Jada Code uses the same configuration system as Goose. Config lives in ~/.config/goose/:

• profiles.yaml — Provider and extension settings
• config.yaml — Global preferences

Environment Variables

Variable	Default	Description
GOOSE_PROVIDER	openrouter	AI provider
GOOSE_MODEL	moonshotai/kimi-k2	Model to use
OPENROUTER_API_KEY	-	OpenRouter API key
GOOSE_HOST	127.0.0.1	Server bind address
GOOSE_PORT	3000	Server listen port
GOOSE_DISABLE_TELEMETRY	1	Telemetry disabled by default

MCP Extensions

Jada Code ships with preconfigured MCP extensions:

• Developer (built-in) — Shell commands, file editing, code analysis
• Memory (built-in) — Persistent storage across sessions
• nc-passwords-mcp — Nextcloud Passwords vault access

Add more extensions via profiles.yaml or the goose configure CLI.

Docker Deployment

Server mode (headless)

docker build -t jada-code .

docker run -d \
  --name jada-code \
  --restart unless-stopped \
  -p 3000:3000 \
  -e GOOSE_PROVIDER=openrouter \
  -e GOOSE_MODEL=moonshotai/kimi-k2 \
  -e OPENROUTER_API_KEY=your-key \
  -e GOOSE_HOST=0.0.0.0 \
  -e GOOSE_PORT=3000 \
  -e GOOSE_DISABLE_TELEMETRY=1 \
  -v jada-code-data:/home/goose/.local/share/goose \
  -v jada-code-config:/home/goose/.config/goose \
  jada-code serve --host 0.0.0.0 --port 3000 --with-builtin developer,computercontroller,memory

Upstream

This is a fork of aaif-goose/goose (Apache 2.0). Goose is part of the Agentic AI Foundation (AAIF) at the Linux Foundation.

To sync with upstream:

git remote add upstream https://github.com/aaif-goose/goose.git
git fetch upstream
git merge upstream/main

License

Apache License 2.0 — see LICENSE for details.