Practical overview
When it fits
Use DeepGit when a repository search needs to preserve detailed project requirements or distinguish between close candidates, especially when keyword searches alone are not sufficient.
Who it is for
- Developers looking for libraries or tools that meet a specific language, hardware or project requirement.
- Researchers who want repository results ranked by fit rather than popularity alone.
- People building MCP-enabled assistant workflows that need GitHub repository research.
Documented uses
- Find repositories matching a natural-language project description.
- Compare candidate repositories against must-haves, preferences and anti-patterns.
- Surface relevant projects using keyword, topic and optional semantic search.
- Inspect source code, tests and manifests when top candidates are contested.
Requirements & limits
The project is open source under the MIT license. Its documented local setup requires Python 3.11+, a GitHub token and credentials for a supported LLM provider.
- The semantic extra is optional; without it, DeepGit still searches by keyword and topic.
- The local semantic index uses LanceDB and fastembed and can retain repository representations across searches.
- Code inspection is reserved for uncertain or contested candidates, so difficult queries may take more LLM calls than typical searches.
- It offers CLI, Python-library, Docker and MCP-server usage paths.
- The README’s benchmark numbers are project-reported results, not an independent evaluation.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Set up DeepGit for a terminal search
Install DeepGit and prepare its documented CLI search prerequisites.
Before you start
- Python 3.11 or later
- A GitHub token with public-repository read access
- Credentials for one supported LLM provider
Steps
Clone the DeepGit repository and change into its directory.
Create and activate a Python virtual environment.
Install the project with the command below; the semantic extra is optional.
pip install -e ".[semantic]"Configure
GITHUB_API_KEYand the chosen LLM provider credentials, then rundeepgit searchwith a natural-language repository request.
Expected outcome
DeepGit returns a repository search result for the supplied request from the terminal.
Guide limitations
- The semantic extra is optional; without it, search uses keyword and topic methods rather than the local semantic index.
- The README does not promise a fixed number of LLM calls; difficult or contested queries may require more work.
Run DeepGit’s MCP server with Docker
Build the documented Docker image and start DeepGit as an MCP server over HTTP.
Before you start
- Docker
- A GitHub token
- LLM provider credentials
Steps
Build the image from the cloned repository with the command below.
docker build -t deepgit .Run the container with port 8080 published and
GITHUB_API_KEYset.Pass the selected LLM provider and its credentials through environment variables or an environment file.
For persistent semantic-index data, mount a volume at
/data.
Expected outcome
The container starts DeepGit’s MCP server over HTTP on port 8080; an optional `/data` volume persists the semantic index across runs.
Guide limitations
- The documented Docker default starts the MCP server; the README shows a separate command form for a one-off CLI search.
- Semantic-index persistence depends on mounting the documented volume.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 14 October 2026
- Original website title
- zamalali/DeepGit · Official homepage title · retained in its source language