Cross-pillar content hub
Agentic AI
Framework reviews at the top. Production stories, implementation notes, and field guides underneath. This hub pulls together the content that matters when you are designing, evaluating, and shipping agent systems.
Framework reviews
17
Hub articles
8
Canonical directory
/agentic-ai/frameworks
Featured frameworks
Start with the directory, not guesswork.
The frameworks block stays at the top of this hub because it is the fastest way to orient new visitors. Reviews live here first, then the supporting articles carry the implementation detail below.
// Featured Frameworks
All 17 frameworks →Open Source
5/5
LangGraph
Graph-based agent orchestration layer from the LangChain team. Models agent execution as a state machine — nodes are steps, edges are transitions, state is persisted. The consensus production framework for stateful agents in 2026.
Best for: Production agent workloads requiring stateful, deterministic execution with persistence, memory, and durable step logic
PythonTypeScript
Open Source
4/5
Google ADK
Google's open-source, code-first framework for building, evaluating, and deploying AI agents. Optimised for Gemini but model-agnostic, with Python, Java, and Go SDKs. Designed for Cloud Run and Vertex AI Agent Engine deployment.
Best for: Teams building on Gemini or deploying to GCP who want a batteries-included framework with built-in observability and Cloud Run / Vertex AI integration
PythonJavaGo
Open Source
4/5
LangChain
The original agent engineering platform for building pipelines, chains, and tool-using agents with Python or TypeScript. The widest ecosystem of integrations with the most documented production gotchas.
Best for: Teams that need maximum integration breadth and can afford to manage the complexity debt
PythonTypeScript
Open Source
4/5
Mastra
TypeScript-native agent framework with built-in workflows, memory, RAG, and MCP support. Designed for full-stack TypeScript teams shipping agents as part of web applications.
Best for: Full-stack TypeScript teams embedding agents inside Next.js or Node.js applications
TypeScript
// Agentic AI articles
Deep Dive
Five attack patterns. One defender's playbook.
How AI Agent Memory Gets Poisoned (And What Operators Can Do)
Agent memory is the layer attackers reach by leaving instructions in places the agent reads. The damage is persistent and crosses sessions. A taxonomy of the five named attack patterns and the operator controls that hold up against them.
Tool Guide
Two production philosophies
PydanticAI vs LangGraph: Two Different Production Bets
PydanticAI brings type-safe, injection-first agents. LangGraph brings stateful graph execution. These are distinct production bets, each shaped by a different philosophy.
Deep Dive
Deep Dive
LangChain vs CrewAI vs AutoGen vs LangGraph: Which AI Agent Framework Should You Actually Use?
Four frameworks, four different production problems. LangChain for breadth, CrewAI for prototyping speed, AutoGen/MAF for enterprise, LangGraph for execution control. Here is how to choose wisely.
Deep Dive
The implementation story behind a machine-readable site.
How We Made AutomationSwitch Readable by AI Agents, and What We Learned
Most writing about the agentic web is theoretical. We skipped the predictions and built it. This is the implementation story: what we built on AutomationSwitch to make the site readable, queryable, and useful to AI agents.
Deep Dive
SKILL.md Files: The Agent Skills Directory
SKILL.md files are the configuration layer for AI agent behavior. How they work, how to write them, and the emerging directory of community-built skills.