Generate Flashcards

Create a markdown file with this structure:

# Flashcards for [Topic Name]

Generated from: [Brief description of source content]

---

## Card 1
Q: What is [concept]?
A: [Clear, concise definition or explanation]

## Card 2
Q: How does [concept] work?
A: [Process explanation with key steps]

## Card 3
Q: What is the difference between [concept A] and [concept B]?
A: [Comparison highlighting key distinctions]

## Card 4
Q: When should you use [technique/tool]?
A: [Use cases and scenarios]

## Card 5
Q: What are the key components of [system]?
A: [Enumeration with brief explanations]

[Continue for remaining cards...]

Quality Guidelines

• Questions should be specific and unambiguous - avoid vague wording
• Answers should be self-contained - don't reference "the above" or assume prior context
• Mix question types: definitions, how-it-works, comparisons, use cases, examples
• Use active voice and clear language
• Include concrete examples in answers when helpful
• Avoid yes/no questions - prefer questions that require explanation
• Test understanding, not memorization - focus on "why" and "how" over "what"

Examples

Example 1: Technical Concept

Q: What is the purpose of the @tool decorator in LangChain? A: The @tool decorator converts a Python function into a tool that can be called by an LLM. It automatically generates a JSON schema from the function's type annotations and docstring, making the function's signature and purpose understandable to language models.

Example 2: Comparison

Q: What is the difference between LangChain and LangGraph? A: LangChain provides building blocks for LLM applications (prompts, chains, tools), while LangGraph extends LangChain with stateful, graph-based workflows. LangGraph is specifically designed for building agents that need persistence, cycles, and complex control flow.

Example 3: Process

Q: How does the ReAct agent pattern work? A: The ReAct (Reasoning-Acting) pattern runs in a loop: the LLM receives a question, decides whether to call tools or provide a final answer, executes any requested tools, receives results, and repeats until no more tools are needed. This enables multi-step problem solving.