Generative AI for Scholarship

Harvard Data Science Initiative (HDSI) & Faculty of Arts and Sciences (FAS)

Session 3 Instructor Guide — Claude Code CLI

Friday, March 6, 2026 · 4:00–5:30 pm · Northwest Building, Room B103

Session Run-of-Show

1. Brief recap of Sessions 1 and 2
   • Session 1: Gemini, NotebookLM, Gems — interactive AI tools
   • Session 2: AI-assisted coding in Colab and Jupyter — cell-by-cell
   • Session 3: Claude Code — autonomous agent that plans and executes
2. Installation and setup
   • Mac users: Mac setup guide
   • Windows users: Windows setup guide (Option 1: VS Code recommended, Option 2: PowerShell CLI)
   • Post the session API key and walk everyone through setup
   • Test with hello.py and plottest.py
   • Expect this to take a significant chunk of time — installation issues are common
3. Launch Claude Code and verify
   • Everyone runs claude and sends a test message
   • Explain the "Auth conflict" warning — it's expected and safe
   • Show basic commands: /help, /cost, /model
4. Thermal data exercise
   • Walk through the exercise together
   • Steps 1–4 (explore, plot, histogram, extremes) go quickly
   • Step 5 (sunset markers) is where it gets interesting — Claude needs to figure out astronomy libraries
   • Step 6 (Fourier analysis) — pause here for discussion about whether results make physical sense
   • Step 7 (critical thinking) — this is the key teaching moment
5. Plan Mode demo
   • Show Shift+Tab to enter Plan Mode
   • Use the ML prediction exercise (Step 10) as the example
   • Walk through the plan review and approval process
6. Wrap-up
   • Show /cost to see session API usage
   • Point to "After the Session" section for getting personal API keys
   • Mention VS Code extension as alternative interface
7. Post-session survey
   • Direct participants to the survey: https://bit.ly/4rlIBCt

Teaching Notes

Common installation issues

• Mac: Some participants may not have Python 3 — check with which python3
• Windows: Python not on PATH (forgot to check the box during install)
• Windows: PowerShell execution policy blocks .ps1 scripts — need Set-ExecutionPolicy
• Both: 403 errors usually mean the API key is wrong or the model is not available
• Both: "Auth conflict" warning is expected and harmless — tell participants to ignore it

Key pedagogical moments

• Step 5 (sunset markers): This shows Claude Code's ability to break complex problems into sub-tasks, install libraries, and combine results
• Step 7 (critical thinking): The Fourier analysis may give unexpected results — the dominant period should be ~1 day (diurnal cycle), but participants should think about whether the data supports this and what other periods appear
• Plan Mode: Emphasize that reviewing the plan is critical — don't just rubber-stamp it. This teaches good habits for working with AI agents

Cost awareness

• The ML exercise (Step 10) is the most expensive step — it involves multiple model calls and iterations
• Have participants run /cost periodically to see cumulative usage
• The workshop API key has a shared quota — if many participants run the ML exercise simultaneously, watch the quota

Solutions and Reference Materials

Discussion Points

• How does working with an autonomous agent differ from cell-by-cell AI assistance?
• When should you trust Claude Code's output vs. verify it manually?
• What are the risks of letting an AI agent read, write, and execute code on your machine?
• How do you evaluate whether a Plan Mode plan is good before approving it?
• What research tasks in your field could benefit from this kind of agentic workflow?