Welcome to the coding agents section.

This series of guides teaches a method for working with coding agents — GitHub Copilot, Claude Code, Cursor, opencode, Codex, Gemini CLI and others — without binding to the syntax of a single tool. The point is not the interface you use, but how you use the agent: context, planning, validation.

The journey is organized into four levels:

• Level 1 — Foundations: what an agent is, the plan→execute→validate work cycle, prompts that work.
• Level 2 — Core practices: context management, repository configuration, choosing between IDE and CLI, sessions and errors, cost and tokens.
• Level 3 — Use cases: the concrete flows — documentation from a codebase, testing and TDD, code review, refactoring, adding features, greenfield from a specification and more.
• Level 4 — Advanced: extending agents with skills and MCP, multi-agent orchestration and spec-driven development.

Output quality depends on method and context, not on the interface. This premise runs through the whole series.

Available guides 📚