Installation¶
Quick Install (macOS / Linux / WSL)¶
This auto-detects the best package manager (pipx > uv > pip) and installs drift in an isolated environment.
Windows (PowerShell):
From PyPI¶
With pipx (recommended for CLI use)¶
pipx installs drift into an isolated environment so it doesn't interfere with your project dependencies:
Or run without installing using uvx:
With Homebrew (macOS / Linux)¶
With Docker¶
This mounts your current directory into the container and runs drift against it. No Python installation required.
With Conda¶
# From conda-forge (when available):
conda install -c conda-forge drift-analyzer
# Or build locally from the repo:
conda build conda.recipe/
conda install --use-local drift-analyzer
From Source¶
Optional Extras¶
# TypeScript/TSX support
pip install -q drift-analyzer[typescript]
# Embedding-based duplicate detection
pip install -q drift-analyzer[embeddings]
# MCP server for IDE integration
pip install -q drift-analyzer[mcp]
# All extras
pip install -q drift-analyzer[all]
Version Pinning¶
For reproducible CI builds, pin to a specific version:
# pip
pip install drift-analyzer
# pipx
pipx install drift-analyzer
# GitHub Action
- uses: mick-gsk/drift@v2.6.0
# Docker
docker run ghcr.io/mick-gsk/drift:v2.6.0 analyze --repo /src
# pre-commit
repos:
- repo: https://github.com/mick-gsk/drift
rev: v2.6.0
Requirements¶
- Python 3.11+ (not needed with Docker)
- Git (for history-based signals)
Editor Integration¶
Drift integrates with AI coding editors via the Model Context Protocol (MCP):
Setup guides: Integrations
A dedicated VS Code Marketplace extension and JetBrains plugin are planned for a future release.