VibeXP

Notes on shared AI knowledge.

How VibeXP works, why it is built the way it is, and what we learn about getting AI tools to remember.

Harness Engineering Explained: Why an AI Agent Is Model + Harness

Harness engineering is everything built around an AI model (the loop, tools, memory, and guardrails) that turns a raw text generator into an agent that reliably gets real work done. A plain-language guide.

VibeXP's MCP Server Is Now Fully MCP-Spec Compliant (OAuth 2.1)

VibeXP's MCP server now implements the official MCP authorization spec as an OAuth 2.1 Resource Server. Spec-compliant clients connect with zero config — paste the URL, sign in, done. No API keys, no secrets in your config.

Get Organized for AI: Give Your Prompts, Context, and Outputs a Home

AI tools multiplied faster than our habits for managing them. Here is how VibeXP turns scattered chats, prompts, and outputs into one organized, reusable knowledge base.

Scattered AI Use Is Quietly Draining Your Efficiency

Lost prompts, forgotten context, and outputs buried in chat history cost more than you think. Here is how scattered AI use compounds — and how to fix it.

Shipping an Agent-Native Blog

How VibeXP builds a file-based, agent-friendly blog where every post is a typed markdown file validated at build time.