Tech notes
AI teams often waste effort on prompts that break in new scenarios, but Anthropic engineers show that upgrading prompts to engineered "Skills"—standardized work units with documentation, scripts, and validation—enables agents to reliably reuse team experience across workflows.
The article shows how mastering just three Linux CLI utilities—grep, awk, and sed—lets engineers filter, analyze, and clean logs quickly, using concrete examples and real‑world cases to locate and resolve the majority of production problems in minutes.
The article analyzes Agent memory design by comparing human short‑term and long‑term memory, explains context‑window management strategies, outlines persistent storage options such as vector databases, relational stores, knowledge graphs and fine‑tuning, and presents a three‑layer architecture with write, retrieval and forgetting mechanisms.
The article analyzes why memory is essential for practical LLM agents, categorizes four memory types, proposes a perception‑judgment‑refinement‑storage pipeline, introduces a three‑dimensional retrieval scoring model, and outlines a three‑layer architecture with reflection, merging, and forgetting mechanisms.
The article explains how confusion between concurrency, latency, and QPS leads to systems that crash under load, using a bank‑counter analogy and a flash‑sale case study to derive a performance formula and practical optimization techniques.
This article explains why embedding models are the foundation of Retrieval‑Augmented Generation, outlines five evaluation dimensions, compares leading open‑source and commercial models, provides a decision tree, practical validation steps, common pitfalls, and future trends to help developers select the most suitable embedding model for their RAG system.
The article explains why RAG storage must be layered, describes the four data types involved, presents a typical three‑layer architecture with vector, content, and metadata stores, and shows how the design evolves with scale, multi‑level indexing, update handling, and tenant isolation.
The article analyzes Claude Code’s common pitfalls—unrun tests, unintended file changes, and bug proliferation—and presents wow‑harness, an open‑source runtime constraint system that uses hooks, an eight‑stage state machine, and tool isolation to enforce strict governance over AI‑generated code, dramatically improving reliability.
This guide walks through deploying a demo application on Kubernetes, then demonstrates how to implement canary releases and blue‑green deployments using Nginx Ingress annotations for header‑based, IP‑based, and weight‑based traffic splitting, complete with YAML manifests and verification commands.
The article explains why RAG systems often retrieve irrelevant results due to a semantic gap between colloquial user questions and formal document language, and presents a four‑layer solution—including query rewriting, HyDE, multi‑query expansion, hierarchical indexing, hybrid search with RRF, rerankers, and embedding fine‑tuning—to systematically close that gap.
