Skip to content
Follow
← all courses

◆ COURSE · 14 lessons (7 available) · 217 min read · free

Claude Code Playbook

For developers in any stack who are new to Claude Code or have used it informally: a first safe session, context and prompting, CLAUDE.md, permissions, skills, subagents, hooks, MCP and plugins, the superpowers cycle from idea to reviewed code, work across several repositories and automation in CI, all on one demo shop of five Python services.

Watch on YouTube

Who it's for

Professional software developers in any stack who are new to Claude Code or have used it informally and want to do real work with it. They read Python but do not need to write it fluently, and they need no Claude Code expertise.

You'll need

  • Proficiency in at least one programming language and comfort reading Python
  • Working knowledge of Git, a terminal and automated tests
  • To run the demo: uv, Git, and Docker for the chapters that need Kafka, plus a Claude account with Claude Code access
  • For the automation chapter: a GitHub account, a copy of the demo repository the reader can push to, and the gh CLI

What you'll be able to do

  • Run, steer, review and undo Claude Code sessions in a real repository
  • Manage what is in the context and carry work across sessions
  • Write prompts and project instructions (CLAUDE.md, rules) that make results predictable
  • Set permissions and a sandbox so an agent has a reduced reach to secrets and to destructive actions, and know what those controls do not cover
  • Extend Claude Code with skills, subagents, hooks, MCP servers and plugins, and share them with a team
  • Take a feature from idea to merged code through brainstorming, a spec, a plan, test-driven implementation and review, with human gates
  • Work across several repositories with shared contracts and an end-to-end check
  • Run Claude Code non-interactively and in CI, and name the safety limits of doing so

Lessons

  1. Lesson 1 · 28 minStart: a first session that changes code safelyInstall Claude Code, check out the demo shop at its first tag, ask orienting questions about one service, make one small change, read the diff before you keep it, and undo it with /rewind.
  2. Lesson 2 · 30 minContext: a focused session and work carried across sessionsSee what fills the context window and measure it with /context, choose between /compact and /clear, keep a session scoped to one service, and carry work into the next session with a handoff.
  3. Lesson 3 · 23 minPrompting for real work: plans, constraints and testsUse plan mode, compare a vague prompt with a precise one, state constraints before any code is written, and define done as passing tests.
  4. Lesson 4 · 36 minCLAUDE.md and memory: instructions a team can keepCut the /init draft down to a CLAUDE.md a team can keep, add a path-scoped rule and @ imports, and see what auto memory stores and why instructions are context, not enforced rules.
  5. Lesson 5 · 40 minPermissions and safety: limiting what the agent can reachPermission modes, the auto-mode classifier and its limits, deny and ask rules that keep .env out of reach, and the OS sandbox around shell commands.
  6. Lesson 6 · 28 minSkills: reusable instructions for you and your teamTurn the handoff prompt into a skill, write a description that lets a skill start on its own, compare it with a published skill from a plugin, and make a new project skill visible in a running session.
  7. Lesson 7 · 32 minSubagents: delegating focused workDelegate read-only work to the built-in Explore and Plan subagents, write a custom reviewer agent, and run work in a worktree that a hook creates next to the service.
  8. Lesson 8 · coming soonHooks: deterministic guardrails around the agentAdd hooks that format after edits, check types when Claude stops and block destructive commands, then find what the blocking hook catches wrongly, what it misses and why a hook that cannot run fails open.
  9. Lesson 9 · coming soonMCP and plugins: tools and team setup as codeConnect an MCP server at local, project and user scope, make a database server read-only at the connection, and package skills, an agent, hooks and a server as a team plugin with its own marketplace.
  10. Lesson 10 · coming soonFrom idea to plan: brainstorming, a spec and a planInstall superpowers at a pinned version and take order cancellation from a broad request through brainstorming, a spec and a plan, with a human review gate after each stage and no code written.
  11. Lesson 11 · coming soonFrom plan to code: subagent-driven execution and reviewExecute the plan with implementer and reviewer subagents in a worktree, read their statuses, run the plan's deliberate break and the final review, then finish the branch and remove the worktree safely.
  12. Lesson 12 · coming soonMany services: orientation in a multi-repository systemFind your way around a system of five repositories: which CLAUDE.md files load from each starting point, how to trace an order across services, and where to keep domain knowledge outside the code.
  13. Lesson 13 · coming soonA feature across services: a coordinated changePlan and build a change across four repositories, add an event without breaking its consumers, make them handle repeats, and gate the result on an end-to-end run against Kafka.
  14. Lesson 14 · coming soonAutomation: claude -p, CI review and output stylesRun Claude Code from scripts with claude -p and JSON output, see why --bare matters in a folder you have not read, set up pull request review in CI, and write a project output style.