A practical breakdown of Claude Code’s agent loop, tool categories, sessions, permissions, and how to work with it effectively.

A hands‑on overview of Easy Dataset—document parsing, smart chunking, QA generation, and built‑in evaluation for fine‑tuning and RAG.

Out of 100,000+ open‑source agent skills, these four are the only ones I genuinely use every day—and why they matter.

A deep, production‑oriented look at LightRAG—graph‑based indexing, dual‑level retrieval, incremental updates, and why it scales better than flat RAG.

From /learn prompts to a full skill pipeline—how to combine search + browsing to build autonomous skills with Gemini 3 models.

A deep dive into Claude Code’s /insights command—how it analyzes your workflow, where it can misread context, and how to turn AI feedback into a production advantage.

A deep read of Cursor’s multi‑agent research—how they moved from single agents to a recursive planner/worker system, and what that means for production pipelines.

A deep dive into CodeWiki—hierarchical decomposition, recursive agents, and multimodal docs for 86K–1.4M LOC repositories.

Opus 4.6 raises the ceiling on coding agents, and Claude Code Agent Teams changes how we orchestrate them. Here’s why the workflow has to evolve.