8 posts tagged AI Prompts. See all posts
An index can turn a slow query into a fast one, or slow down every insert on the table. In this article, we explain database indexing in SQL Server, PostgreSQL, and MySQL: eleven index types, when to use each one, when to avoid them, and how to find missing and unused indexes.
This week: Anthropic redesigns code review around evidence and risk; LinkedIn combines generated data documentation with owner-maintained knowledge; Uber reduces translation-cache memory without introducing network calls; and Stripe discovers why more training data can make fraud detection worse.
Weekend Reading #98: AI agent terminology, Uber’s MCP Gateway, Zalando’s delegated identity broker, and Meta’s petabit-class subsea cable design.
30 AI terms in two or three plain sentences each, placed on one diagram: what the model sees, what it can call, and what runs around it. MCP, skills, hooks, subagents, RAG, workflows, and the pairs people mix up.
Weekend Reading: A weekly roundup of interesting Software Architecture and Programming articles from tech companies. Find fresh ideas and insights every weekend.
Once an AI agent can push to Git, prompt injection becomes an authorization problem. Six things to build before agents get write access: a scoped identity, a policy outside the model, a sandbox, MCP review, advisory CI, and traces.
EF Core makes the slow path look exactly like the fast path. A scan, an N+1 loop, and a Cartesian join all read like clean LINQ. This guide walks through 18 concrete mistakes across querying, related data, change tracking, indexes, plan caching, and context overhead, with the fix for each and a ready-to-use rule for your AI coding assistant.
AI speeds up coding, but shifts pressure to review, testing, security, maintainability, and CI. This article examines seven failure modes of AI-assisted development and where engineering teams should respond.