NOTES / PART 2 · BUILDING AGENTS

07Frameworks and SDKs

Libraries for building agents, compared, and when to skip them.

THIS SECTION COVERS

Agent SDKs and orchestration frameworks, how each maps to the patterns and harness pieces, trade-offs, building directly on model APIs.

Related: Harnesses · Agentic design patterns

TOPICS · 8/8 WRITTEN

  1. 07.1Building without a frameworkAn agent is a loop around a model API. Start with plain API calls, learn what a framework would add, and adopt one only when it earns its abstraction.
  2. 07.2Choosing a frameworkHow to compare agent frameworks: what layer each occupies, which problems each is built for, and a short list of questions that decide the choice.
  3. 07.3LangGraphA low-level runtime for stateful agents: graphs of nodes and edges over shared state, with checkpointing, interrupts and streaming built in.
  4. 07.4OpenAI Agents SDKA lightweight SDK built on a few primitives: agents, handoffs, guardrails, sessions and tracing, with a managed run loop.
  5. 07.5Claude Agent SDKThe agent loop, built-in tools and context management that power Claude Code, packaged as a Python and TypeScript library you embed in your own app.
  6. 07.6Google ADKGoogle's Agent Development Kit: LLM agents plus Sequential, Parallel and Loop workflow agents, composed into hierarchies, across Python, TypeScript, Go and Java.
  7. 07.7CrewAIA framework built around agents with roles, crews that collaborate on tasks, and flows that give you event-driven control with state and persistence.
  8. 07.8Pydantic AIA Python agent framework from the Pydantic team: agents generic over dependencies and output types, validated structured output, and typed dependency injection for testing.