Level 2, intermediate. Your first weeks with Claude Code or Codex: how an agent uses tools and files, when to split work across sub-agents, and how to keep it on the rails.
Updated July 5, 2026
This Level 2 guide is for people who already use AI chat and are trying a coding agent like Claude Code or Codex for the first time. You will learn how agents use tools, project context, sub-agents, MCP connectors, and loops so you can let an agent work in your files without losing control.
Who this is for: You are trying Claude Code or Codex for the first time and want your agent to use tools, edit files, and delegate to sub-agents well.
A coding agent is different from chat because it can keep a goal in view while it reads your project, uses tools, checks results, and adjusts its next step.
Read the deep diveBefore asking a coding agent to change files, give it a small, durable map of your project so it can find the right context, respect boundaries, and use its tools with less guesswork.
Read the deep diveA reusable skill is the point where a working prompt becomes a small operating procedure your coding agent can run consistently, check, and improve.
Read the deep diveUse sub-agents as disposable specialists, not extra decision-makers: the main agent keeps the plan, owns the merge, and proves the result works.
Read the deep diveMCP turns a coding agent from a smart chat box into a controlled operator for your real apps, so the first skill is deciding exactly what it may see, do, and remember.
Read the deep diveA safe agent loop is not a bigger prompt; it is a bounded system that repeats a known procedure until an observable stop condition says the work is done.
Read the deep dive