Roadmap

The Skill Tree

Path	For whom	Levels
Beginner	No AI SDK experience	All levels in order (1–9)
Advanced	Already have AI SDK experience	3, 5, 6, 8, 9
Agents & MCP	Focus on agents and tools	1, 3, 4, 8, 9
Production	Focus on quality and robustness	1, 5, 6, 7, 9

Level 1 — AI SDK Basics (6 Challenges) Generate text, stream, structured output, system prompts
Level 2 — LLM Fundamentals (4 Challenges) Tokens, context windows, usage tracking, prompt caching
Level 3 — Agents & MCP (5 Challenges) Tool calling, MCP via stdio/HTTP, tool loop agent
Level 4 — Persistence (4 Challenges) Callbacks, chat IDs, database persistence, message validation
Level 5 — Context Engineering (5 Challenges) Prompt templates, exemplars, RAG, chain of thought
Level 6 — Evals (5 Challenges) Evalite, deterministic evals, LLM-as-judge, Langfuse
Level 7 — Streaming (4 Challenges) Custom data parts, message metadata, error handling
Level 8 — Workflows (4 Challenges) Multi-step workflows, custom loops, frontend streaming
Level 9 — Advanced Patterns (4 Challenges) Guardrails, model router, output comparison, research workflow