obsidian-copilot

What is Obsidian-Copilot?

Obsidian-Copilot is a project designed to function as a virtual assistant for writing and thinking, integrated with the Obsidian note-taking platform. Imagine having a copilot that helps structure your writing process by retrieving relevant notes and documents to draft sections of your work. This tool leverages a method known as retrieval-augmented generation, providing enhancements in other functionalities as well. For example, it can help you reflect on your past week by reviewing your daily journal entries.

Currently, Obsidian-Copilot benefits users by:

Assisting in drafting sections based on existing notes.
Offering weekly reflections using journal entries.

For those interested in the more technical aspects of how it operates, further details can be found in Obsidian-Copilot: A Prototype Assistant for Writing & Thinking.

Getting Started with Obsidian-Copilot

To get started with Obsidian-Copilot, begin by cloning the repository and updating paths to your Obsidian vault and huggingface hub cache.

Clone the repository:

git clone https://github.com/eugeneyan/obsidian-copilot.git

Update your terminal configuration files (such as ~/.bashrc or ~/.zshrc) to include the OBSIDIAN_PATH and TRANSFORMER_CACHE paths, followed by sourcing it. Ensure your path ends with a trailing slash:
```
export OBSIDIAN_PATH=/path/to/obsidian-vault/
export TRANSFORMER_CACHE=/path/to/.cache/huggingface/hub
```
Create the huggingface hub cache directory if it doesn't exist:
```
mkdir -p $TRANSFORMER_CACHE
```

Build the OpenSearch and semantic indices by running:

make build    # Build the docker image
make opensearch  # Start the opensearch container
make build-artifacts  # Build your artifacts

Start the retrieval app after stopping the opensearch using CTRL + C:
```
make run  # Start the retrieval app
```
Install the copilot plugin, enable it in the community plugin settings, and update the API key:
```
make install-plugin
```

How Does It Work?

The copilot's functionality is designed to enhance productivity and creativity by responding to section headers input by the user. Upon entering a section header, the copilot retrieves relevant documents or snippets from your Obsidian vault using:

Keyword search: Conducted through OpenSearch.
Semantic search: Conducted through semantic retrieval.

This retrieved context provides the basis for generating paragraphs for your section. The information utilized is then presented in a new tab for reference.

Contribution and Development

To contribute to the project, you can install pre-commit hooks by executing:

pip install pre-commit && pre-commit install

This should be done at the root of the repository to streamline code contributions and ensure consistency.

Future Developments

The project roadmap includes plans to:

Add support for using the Anthrophic Claude with a 100k context.
Evaluate the possibility of sending entire documents rather than just chunks for processing.

Through continuous development, Obsidian-Copilot aims to make the writing and thinking process more seamless and intuitive for its users.