Getting Started

Overview

SourceSage scans a repository and writes an AI-friendly Markdown documentation artifact to .SourceSageAssets/Repository_summary.md.

Try it once with `uvx`

bash

uvx --refresh sourcesage --help
uvx --refresh sourcesage --repo /path/to/repository

Run from source

bash

git clone https://github.com/Sunwood-ai-labs/SourceSage.git
cd SourceSage
uv sync
uv run sage --help
uv run sage --repo .

This creates .SourceSageignore when missing and writes the primary documentation artifact under .SourceSageAssets/.

Start with a lighter first pass

bash

uv run sage --repo . --lite

Use --lite first when ignore rules are not tuned yet. It keeps the tree, Git info, stats, and root README files without expanding every file into the summary.

Analyze another repository from this checkout

bash

uv run --directory D:\Prj\SourceSage sage --repo D:\Prj\SourceSage\example -o D:\Prj\SourceSage\.tmp-docs-check\example

Replace those paths with your own SourceSage checkout and target repository.

Generate Japanese documentation into a custom directory

bash

uv run sage --repo . -l ja -o ./out

Verify the repository locally

bash

uv run pytest -q

Next steps

Review flags in the CLI Reference.
Inspect the generated files in the Output Guide.
If something looks wrong, start with Troubleshooting.

Getting Started ​

Overview ​

Try it once with uvx ​

Run from source ​

Start with a lighter first pass ​

Analyze another repository from this checkout ​

Generate Japanese documentation into a custom directory ​

Verify the repository locally ​

Next steps ​

Getting Started

Overview

Try it once with `uvx`

Run from source

Start with a lighter first pass

Analyze another repository from this checkout

Generate Japanese documentation into a custom directory

Verify the repository locally

Next steps