Installation

GooGo is currently installed directly from GitHub.

npm installation will be added later.

1. Clone the repository

git clone https://github.com/Forzun/GooGo.git

Enter the project:

cd GooGo

2. Install dependencies

GooGo uses Bun.

bun install

3. Install Ollama

Install Ollama for your operating system, then make sure the Ollama service is running.

You can verify it with:

ollama list

Download a Chat Model

GooGo lets you use models installed in Ollama.

For example:

ollama pull qwen3.6:27b

You can use another compatible Ollama model if you prefer.

Check your installed models:

ollama list

When GooGo starts, you can select from your locally installed models.

Download the Memory Embedding Model

GooGo's long-term memory system uses a separate embedding model.

The embedding model does not generate chat responses.

It converts memories and questions into numerical vectors so GooGo can perform semantic search and find relevant memories.

Here are some recommended embedding models list

- ollama pull Qwen3-Embedding-8B
- ollama pull Qwen3-Embedding-4B
- ollama pull mxbai-embed-large

For more check here: https://ollama.com/search?c=embedding

Download the embedding model:

ollama pull nomic-embed-text

Make sure it appears in:

ollama list

this is just how rag work

The chat model and embedding model have different jobs:

Chat Model
    │
    └── Thinks, answers questions and works with code

nomic-embed-text
    │
    └── Finds semantically relevant memories

Running GooGo

During development, run:

bun link

now simply you need to run goo command to use it

goo

GooGo will detect your locally installed Ollama models and ask you to select one.

? Select a model

● qwen3.6:27b
● llama3.2:8b
● ...

After selecting a model, you can start chatting.

Memory Architecture

GooGo follows an important principle:

Markdown is the source of truth. SQLite is the search index.

Your permanent memory lives inside Markdown files.

Alt text for the image

This means the SQLite database is disposable.

If the vector database is deleted, the original Markdown memories still exist and can eventually be re-indexed.

-->

Memory Location

By default, GooGo stores its data inside:

~/.goo/

The structure looks roughly like:

~/.goo/
│
├── config.json
│
├── vector.db
│
└── vault/
    │
    ├── Daily/
    ├── Memory/
    ├── People/
    ├── Preferences/
    └── Projects/

The vault directory contains the human-readable Markdown memory.

The vector.db file contains the SQLite search index.

Open GooGo Memory in Obsidian

GooGo's memory vault is compatible with Obsidian.

To explore your memory:

Open Obsidian.
Choose Open folder as vault.
Select:

~/.goo/vault

For example:

/home/your-username/.goo/vault

Once opened, you can browse GooGo's Markdown memory directly inside Obsidian.

View Memory as a Graph

After opening:

~/.goo/vault

as an Obsidian vault, open Graph View.

As GooGo's memory system grows and notes begin using Obsidian links such as:

The long-term goal is to combine:

Semantic vector search
Markdown knowledge
Obsidian links
Tags
Projects
Preferences
People
Recency

This allows GooGo's memory to remain both machine-searchable and human-readable.

Memory Privacy

GooGo's memory is stored locally on your machine.

Your vault lives inside:

~/.goo/vault

Image Description

You can:

Read your memories
Edit them
Delete them
Back them up
Version-control them
Open them in Obsidian

The goal is simple:

Your AI memory should belong to you.

Now you done goo and enjoy after this its all about upcoming feature

Contributing

GooGo is open source and experimental.

If you find a bug, have an idea, or want to improve the project, contributions are welcome.