Source: Agent Development Kits (Loic's Notebook)

The owner's tutorial comparing Google ADK, AWS Strands Agents, Bedrock AgentCore and LangGraph.

Contents

Summary

Frames an agent development kit as a category that packages eight production concerns (loop, tools, orchestration, memory, streaming, guardrails, evaluation, deployment) and compares four options in a problem/solution/limits format. Google ADK is presented as a batteries-included framework with both LLM agents and deterministic workflow agents.

Key claims

  • The key axis between kits is who decides the route: your code (workflow agents) or the model (delegation).
  • Google ADK primitives (1.x era): Agent/LlmAgent, SequentialAgent, ParallelAgent, LoopAgent, sub-agents, Runner, Session/State/Memory, tools, callbacks.
  • ADK limits: young and fast-moving with breaking changes; model-agnostic via LiteLLM but smoothest with Gemini and Vertex AI; smaller community than LangChain.
  • Authoring frameworks (ADK, Strands) compose with managed runtimes such as Bedrock AgentCore.

Feeds

Suggest changes →