PPBDS devcontainers

Container infrastructure for the PPBDS GitHub organization. This repo publishes three Docker images to the GitHub Container Registry (GHCR); other PPBDS repos consume them via short devcontainer.json files.

Images

Image	Audience	Built from
ghcr.io/ppbds/devcontainers/base	shared substrate (don't consume directly)	rocker-org/devcontainer/tidyverse
ghcr.io/ppbds/devcontainers/dev	developing PPBDS R packages	base
ghcr.io/ppbds/devcontainers/student	taking the data science course	base

Consuming

In a downstream repo's .devcontainer/devcontainer.json:

{
  "image": "ghcr.io/ppbds/devcontainers/dev:X.Y.Z"
}

Replace X.Y.Z with the latest tag from the releases page. Pin to a specific semver tag for stability — :latest floats with main and is fine for development but not for student-facing repos.

Tag families

• :latest — most recent successful build from main. Convenient, not stable.
• :X.Y.Z and :X.Y — semver tags from a GitHub release on this repo. The canonical stable pin.
• :<semester> (e.g. :fa25) — student only. A moving channel for a single semester, applied manually by retagging a tested release.

AI coding assistants

The base image (and therefore dev and student) ships three AI coding CLIs so that students and developers can pick the model that fits their cost and quality needs:

• claude — Claude Code. Anthropic's CLI, single-provider, uses Claude models. Highest quality, highest cost.
• gemini — Gemini CLI. Google's CLI, free tier available.
• aider — Aider. Multi-provider. Point it at DeepSeek, OpenRouter, OpenAI, Anthropic, or any OpenAI-compatible endpoint. The cost-flexible option.

The images themselves ship no credentials. Each CLI is inert until the relevant API key is present in the environment. Students supply their own keys via Codespaces user secrets (instructions below).

Setting up API keys (per-user, one-time)

Configure each key as a personal Codespaces secret so it appears as an environment variable in every Codespace you launch — no need to re-enter it for each new Codespace.

1. Go to https://github.com/settings/codespaces.

2. Under Codespaces secrets, click New secret.

3. Add the secret name and value for each tool you plan to use:

   Secret name	Used by	Where to get it
   ANTHROPIC_API_KEY	claude	https://console.anthropic.com
   GOOGLE_API_KEY	gemini (paid tier; free tier uses OAuth)	https://aistudio.google.com/apikey
   DEEPSEEK_API_KEY	aider → DeepSeek directly	https://platform.deepseek.com
   OPENROUTER_API_KEY	aider → any model via OpenRouter	https://openrouter.ai/keys

4. Under Repository access, grant access to the repos you launch Codespaces from (typically your codespace-starter-derived repo).

5. Save. The next Codespace you launch will have those env vars available, and the CLIs will pick them up automatically.

You only need to set up the keys for the tools you actually plan to use. Most students set one of (DEEPSEEK_API_KEY or OPENROUTER_API_KEY) plus optionally ANTHROPIC_API_KEY for higher-quality work when needed.

Using cheaper models via Aider

For cost-sensitive work, point Aider at a cheap model:

# DeepSeek directly — typically ~$0.02–0.05 per 50K-token coding session
export DEEPSEEK_API_KEY=...   # already set via Codespaces secret
aider --model deepseek/deepseek-chat

# Or via OpenRouter — one key, hundreds of models you can swap between
export OPENROUTER_API_KEY=... # already set via Codespaces secret
aider --model openrouter/deepseek/deepseek-chat
aider --model openrouter/qwen/qwen-3-coder
aider --model openrouter/google/gemini-2.5-flash

For frontier-quality answers on hard problems, use claude directly (charged at Anthropic's rates) or aider --model anthropic/claude-sonnet-4-6.

Architecture and design rationale

See CLAUDE.md. It documents what goes in each image, why, and the conventions to follow when changing them.