Concepts
The questions I kept answering twice
These six pieces exist because the same questions kept arriving in my inbox and in issues. What MCP actually is once you strip the marketing, when stdio beats HTTP, how the memory architectures really differ, and what agent orchestration looks like when you have run it in production. Each one is written from building the thing, not from reading about it. None of them sells anything.
- protocol5 min
What is MCP, actually?
Model Context Protocol explained without the marketing. What it solves, what it doesn't, and when stdio beats HTTP.
Read → - architecture4 min
Stdio vs HTTP for MCP
Transport choice changes everything. Latency, security model, deployment story. A practical decision tree.
Read → - memory6 min
Memory architectures compared
Local SQLite + FTS5, knowledge graph, vector embeddings, hybrid. Which fits which agent shape?
Read → - protocol3 min
Agent-to-Agent protocol v1.0 RC
What an agent-card.json buys you, what it doesn't, and how it relates to MCP and WebMCP.
Read → - browser3 min
WebMCP for browser agents
W3C CG Draft, Chrome 146+. The proposal to give web pages a structured tool surface for in-browser agents.
Read → - agents5 min
Agent orchestration patterns
Single agent, sequential pipeline, parallel fleet, judge-arbitrated. Trade-offs from running each in production.
Read →