Skip to main content
ToolDix logoToolDixFind, compare, and use tools
Sign In
Browse 20 learning paths

Deep-dive course

Claude Code Tutorial: From First Command to Custom Workflows

20 lessons, roughly 325 minutes end to end. Start from what Claude Code is and build up to extending it with custom tools, subagents, and hooks -- with diagrams, worked examples, and production-ready patterns at every step.

Start the course

Course outline

  1. 1What Claude Code Is: Architecture and Core DistinctionClaude Code is an agentic coding tool that reads your codebase, edits files, and runs commands across a multi-step decision loop. This lesson covers its architecture, the three execution surfaces, and what makes it fundamentally different from a chat assistant or IDE copilot.Beginner16 min
  2. 2Chat Assistant vs. Agentic CLI: Architecture and WorkflowSame model, different architecture. Chat assistants see only what you paste; Claude Code reads your entire codebase and decides its own next steps. This lesson explores the mechanical differences and when to use each.Beginner15 min
  3. 3The Agent Loop: How Claude Code Executes One TurnInside Claude Code's decision loop -- tool_use messages, tool_result structures, context accumulation across turns, when the agent decides to stop, and how the loop handles errors and constraints.Intermediate18 min
  4. 4Slash Commands and Custom Skills: Syntax, Frontmatter, and InvocationSlash commands extend Claude Code with custom workflows. This lesson covers the exact file format (.claude/commands/*.md), frontmatter fields, variable substitution patterns, bundled commands, and how custom commands integrate with the agent loop.Intermediate17 min
  5. 5Permission Modes: Control Approval FrequencyClaude Code supports six permission modes that control how often you approve actions. This lesson covers each mode's exact behavior, when to switch, and how to set modes at startup or as a default.Intermediate14 min
  6. 6Permission Rule Syntax: Allow, Ask, and Deny RulesFine-grained permission rules control exactly which tools Claude Code can use. This lesson covers rule syntax, pattern matching, rule precedence, and how sandbox isolation interacts with permission rules.Intermediate15 min
  7. 7Security Case Studies: Real Threat ScenariosFour realistic security scenarios show how Claude Code's permission system catches or fails to catch attacks. Each case includes the threat, the exact attack attempt, and the rule or sandbox configuration that prevents it.Advanced16 min
  8. 8CLAUDE.md Fundamentals: Project Memory and ContextCLAUDE.md is a standing instruction file that gives Claude Code persistent context about your project. This lesson covers the file format, exact load order, @import syntax, and how context is injected at session start.Intermediate14 min
  9. 9CLAUDE.md Advanced Patterns: Monorepos, Hierarchies, and Auto MemoryAdvanced CLAUDE.md techniques for large codebases, hierarchical structures, the .claude directory, auto memory management, and patterns for keeping instructions useful over time.Advanced15 min
  10. 10The Complete Hook Event LifecycleClaude Code supports 30+ hook event types fired at precise moments in the agent loop. Learn the exact timing, input schemas, and event-specific fields for each lifecycle phase.Intermediate18 min
  11. 11Writing Hooks: The Contract and RegistrationHooks are shell scripts that run on stdin/stdout. Learn the exact input/output contract, how to register hooks in settings, and how exit codes control behavior.Advanced19 min
  12. 12Hook Recipes: 4 Real-World AutomationsFour complete hook implementations: auto-formatting after edits, blocking risky paths, logging every command, and desktop notifications on completion.Intermediate17 min
  13. 13MCP Fundamentals: Architecture and Tool InvocationThe Model Context Protocol connects Claude Code to external tools. Learn the client-server model, how tools are discovered and named, and how results flow back into the agent loop.Intermediate16 min
  14. 14Configuring MCP Servers: `.mcp.json` Schema and ScopesExact configuration: server types (stdio, HTTP, SSE), all required/optional fields, environment variables, per-scope settings, and managed MCP for organizations.Advanced20 min
  15. 15Subagents: Isolation, Specialization, and Parallel WorkSubagents are specialized AI assistants that run in isolated contexts with scoped tools and custom prompts. Learn when to delegate tasks, the exact subagent file structure, and how discovery matching works.Intermediate15 min
  16. 16Building Custom Subagents: File Format, Discovery, and LifecycleMaster the exact syntax for defining custom subagents, how Claude discovers and matches them, and how to build a production subagent from scratch with proper error handling and output structure.Intermediate16 min
  17. 17Orchestrating Multiple Subagents: Parallelism, Pipelines, and Trade-offsMaster patterns for coordinating multiple subagents: parallel dispatch, sequential pipelines, failure handling, and cost management. Learn when to use agent teams instead.Advanced17 min
  18. 18Skills: Packaging Reusable ExpertiseSkills are reusable, packaged workflows that Claude Code can call. They encapsulate expertise (how to deploy, how to debug, how to write tests) so it's available in every session.Intermediate10 min
  19. 19The Claude Agent SDK: Building Agents ProgrammaticallyThe Claude Agent SDK exposes the same agentic engine that powers Claude Code as a Python and TypeScript library. Learn the core query() API, how it relates to Claude Code, and how to build production agents with full control over tools, hooks, and MCP servers.Advanced18 min
  20. 20Best Practices Checklist: Production Claude Code and SDKA rigorous, concrete best-practices checklist for using Claude Code in production teams. Covers CLAUDE.md hygiene, permission scoping, review discipline, cost management, and orchestration without over-complexity.Advanced19 min

What this course covers

Every lesson includes worked examples and concrete scenarios. Several lessons include original ToolDix diagrams of the permission boundary, the core workflow loop, CLAUDE.md context loading, subagent delegation, hooks lifecycle, MCP architecture, and best practices. Each lesson cites primary sources -- official Claude Code documentation, Anthropic engineering posts, and current provider documentation -- so you can go deeper on any single topic.