Zero to Container: Syncing VS Code with Docker and uv

In modern AI development, consistency is everything. To build a robust AI Analyst agent, you need an environment that is isolated but accessible. By combining uv, Docker, and Bind Mounts, you can write code in your local VS Code while it executes instantly inside a high-performance container.

Phase 0: Getting uv on Your Machine

Before we containerize everything, you need uv installed locally to initialize your projects.

Step 1: Initialize the Local Project

Create your project folder and use uv to seed the basic Python structure.

Note: Rename the default hello.py to main.py to follow standard naming conventions.

Step 2: Create the Dockerfile

This file defines your environment. We removed the --frozen flag to ensure uv can generate a lockfile during the first build.

FROM python:3.12-slim

# Install uv from the official binary
COPY --from=ghcr.io/astral-sh/uv:latest /uv /uvx /bin/

# Set performance and sync settings
ENV UV_LINK_MODE=copy \
    UV_COMPILE_BYTECODE=1 \
    PYTHONUNBUFFERED=1

WORKDIR /app

# Install dependencies first (better caching)
COPY pyproject.toml uv.lock* ./
RUN uv sync --no-install-project

# Copy project files
COPY . .
RUN uv sync

Step 3: Create docker-compose.yml

This handles the volume mapping and keeps the container running with tail -f /dev/null.

services:
  agent:
    build: .
    container_name: ai_analyst_dev
    volumes:
      - .:/app
    env_file:
      - .env
    environment:
      - UV_LINK_MODE=copy
    tty: true
    stdin_open: true
    command: tail -f /dev/null

Step 4: Configuration & Secrets

Create a .dockerignore file to prevent local junk from entering your container:

Finally, create a .env file in your local folder to store your API keys: