Day 1 · Block 2

E02 - Prompt anatomy in runnable Python

• Familiarize yourself with LangChain https://docs.langchain.com/ and OpenRouter https://docs.openrouter.ai/ by browsing their docs and examples.
• What all can you do with LangChain?
• What are some of the most powerful models on OpenRouter? What are the "best value" models? What do they say about privacy and data usage?
• Run `python exercises/02/prompt_lab.py`.
• Inspect `TODO-STUDENT` prompt instruction and modify strictness once.
• Re-run and compare extraction output quality.
• Verify the resulting list excludes fictional places.

Inputs

• exercises/02/prompt_lab.py
• Noisy country paragraph
• One TODO-STUDENT prompt variant

Deliverable

one runnable extraction output + one verification note + one TODO experiment note.

Target

Familiarize yourself with LangChain https://docs.langchain.com/ and OpenRouter https://docs.openrouter.ai/ by browsing their docs and examples.

Submission

• Use the live launcher: Live Exercises
• Direct prefilled form: E02 submission link

Checklist

• Make one explicit design decision.
• Include one verification check.
• State one limitation or risk.

Common failure modes

• Output format drifts from strict list.
• Typos fixed inconsistently or fictional places included.

Extension task (optional)

Optional multi-agent blackjack demo (Dealer, Gambler, Referee) using explicit standard rules: card values, initial deal, hit/stand, bust, dealer hits to 17+, blackjack precedence, and push on equal totals.