neogit 0.15.0

Git-like tool backed by Neo4j

neogit

A Git-like tool for filesystems, backed by a Neo4j graph and pluggable object storage.

Neogit takes content-addressed Merkle-tree snapshots of a directory tree and stores them in two places:

• Neo4j — the graph: commits, branches, trees, blobs, and their relationships
• Object storage — the bytes: file contents addressed by their SHA-1 (local filesystem, MinIO, or S3 via Apache Libcloud)

This split makes filesystem state queryable as a graph (Cypher over commits, diff trees, walk history) while keeping file contents in cheap blob storage.

Where it's used

• CLI tool — capture and diff filesystem snapshots from the command line
• Python library — embed the Merkle model and Neo4j layer in your own pipeline. For example, OSWatcher uses neogit as the storage foundation, and downstream projects build analysis plugins (filetype detection, symbol extraction, syscall tracing, …) on top of the Commit / PluginRun graph

Quickstart

Requirements: Python 3.10+, Docker, and Git. Neogit uses local object storage by default, so the minimal setup only needs Neo4j.

pipx install neogit
# or: python -m pip install neogit

# Start a local Neo4j database for the demo. Auth disabled is for local testing only.
docker run --rm --name neogit-neo4j \
  -p 7474:7474 -p 7687:7687 \
  -e NEO4J_AUTH=none \
  neo4j:5.26

In another terminal, snapshot a real project checkout:

git clone --depth 1 https://github.com/psf/requests.git neogit-demo-root

neogit init
neogit commit first-snapshot -r ./neogit-demo-root

Open the Neo4j browser at http://localhost:7474 and run:

MATCH (c:Commit)-[r]->(t:Tree) RETURN c, r, t LIMIT 25

CLI overview

neogit init                                    # initialize database constraints
neogit commit <name> -r <path>                 # snapshot a directory on the default branch
neogit diff <old_hash> <new_hash>              # compare two filesystem snapshots
neogit branch <name> <commit_hash>             # create a branch pointing at a commit hash

See docs/reference/cli.md for the full reference.

Use as a library

from pathlib import Path
from neogit.service import Neogit

git = Neogit()
git.init()
commit_hash = git.commit("snapshot-1", Path("/path/to/capture"))

The graph model (Commit, Branch, Tree, Blob, PluginRun) is exposed under neogit.model for downstream tools that want to attach their own nodes — see docs/reference/data-model.md.

Documentation

Full documentation lives under docs/ and follows the Divio framework:

• Tutorial — your first snapshot in 5 minutes
• How-to guides — recipes for specific tasks (MinIO, S3, diffs, embedding the library)
• Reference — CLI flags, config keys, data model
• Explanation — design rationale, Merkle layout, why Neo4j

To preview the docs locally:

poetry install --with docs       # one-time, installs mkdocs into the venv
poetry run poe docs_serve        # equivalent to: poetry run mkdocs serve

Development

poetry install
poetry run poe ccode      # fmt + lint + type-check
poetry run poe test       # full test suite

See docs/how-to/contributing.md for the full dev workflow.

License

Licensed under the Apache License 2.0. You're free to use, modify, and distribute neogit, including commercially, provided you preserve the copyright and license notices (see NOTICE).