VibeGit MCP Server

A Model Context Protocol (MCP) server for logging and analyzing AI assistant conversations.

Prerequisites

You need only two steps to get started:

Step 1: Installation

pip install vibegit-mcp

Step 2: Configuration

Once installed, you can configure the MCP configuration file to enable the VibeGit MCP server. Assuming you are using VSCode, you can add a mcp.json file in the .vscode/ directory of your project with the following content:

{
  "servers": {
    "vibegit": {
      "type": "stdio",
      "command": "vibegit-mcp"
    }
  }
}

Related MCP server: Memory MCP

Usage

After configuring the MCP server, you can start your AI Coding Agent in VSCode. The VibeGit MCP server will automatically log all conversation rounds to the .vibe/ directory in your project root.

Features

• Log complete conversation rounds between users and AI assistants

• Track file operations and tool usage

All the logs and data are stored in the .vibe/ directory under the project root. The directory structure is as follows:

.vibe/
├── rounds/
│   ├── 2023-03/
│   │   ├── round-1.json
│   │   ├── round-2.json
│   ├── 2023-04/
│   │   ├── round-3.json
│   │   ├── round-4.json
├── index.jsonl
├── sessions/
│   ├── session-1.json
│   ├── session-2.json

Each round-*.json file contains detailed information about a single conversation round, including user inputs, AI responses, and any file operations and tool usage performed. The index.jsonl file provides a quick reference to all rounds, and the sessions/ directory contains session metadata. Each session contains the consecutive rounds of conversations.

Building and Publishing (For Maintainers)

This package uses modern Python packaging with pyproject.toml.

Prerequisites

Install build tools:

pip install build twine

Set up PyPI credentials in ~/.pypirc:

[distutils]
index-servers =
    pypi
    testpypi

[pypi]
repository = https://upload.pypi.org/legacy/
username = __token__
password = # your PyPI API token (pypi-...)

[testpypi]
repository = https://test.pypi.org/legacy/
username = __token__
password = # your TestPyPI API token (pypi-...)

Release Process

1. Update version in pyproject.toml:

   version = "x.y.z"  # Increment as needed

2. Clean previous builds:

   rm -rf dist/ build/ *.egg-info

3. Build the package:

   python -m build

4. Test upload to TestPyPI (optional but recommended):

   python -m twine upload --repository testpypi dist/*

5. Test installation from TestPyPI:

   pip install --index-url https://test.pypi.org/simple/ --extra-index-url https://pypi.org/simple/ vibegit-mcp==x.y.z

6. Upload to PyPI:

   python -m twine upload dist/*

Notes

• Always test with TestPyPI first before publishing to PyPI

• Make sure to increment the version number for each release

• The package uses pyproject.toml for modern Python packaging standards

• Clean the dist/ directory before building new releases

License

MIT License