/advisor fableAn advisor model consulted at key moments: plans, reviews, architecture decisions. When to turn it on, when to switch it off.
Module 4Updated for Claude Fable 5.1 · September 2026
Master Claude Code, from your first slash command to CI
The hundred slash commands, CLAUDE.md, skills, permissions, hooks, subagents, MCP, plugins, GitHub Actions and the Agent SDK — learned on a real team project, not on toy examples.
- 8 hrs · 16 modules
- 100 slash commands
- 40 exam questions
- Verifiable certificate
❯
Claude Fable 5.1 just shipped
The course is already up to date
The content follows the official Claude Code documentation as of September 2026. Every command cited really exists; when behavior depends on the version, the course says so.
/effort low → ultracodeSeven effort levels to match reasoning depth to the task, and what each one does to your bill.
Module 4/fastFast mode for small, repetitive, low-latency tasks, and the cases where it costs you more than it saves.
Module 4/fork · /subtaskHand long-running work to a background session or to a subagent that inherits the whole conversation.
Modules 4 and 10/context · /autocompactRead context-window usage, set the compaction threshold and keep multi-hour sessions going.
Module 4Skills · hooks · MCP · pluginsCustom commands turned into skills, deterministic hooks, MCP servers and packaging as a team plugin.
Modules 6, 9, 11, 12An engineering tool, not a chatbot
What you’ll be able to do when you’re done
- Know every built-in slash command, its aliases and its arguments, and when to type it.
- Write a CLAUDE.md and targeted rules that hold up over time.
- Build your own commands as skills, with arguments and execution inside a subagent.
- Lock down what needs locking with permissions, a sandbox and deterministic hooks.
- Parallelize with subagents, connect your tools through MCP, distribute as a plugin.
- Run Claude Code without a UI:
claude -p, GitHub Actions, Agent SDK. - Keep costs under control, prevent prompt injection and deploy across an organization.
Kiosque, the running example: a FastAPI API inherited from a contractor, 41% test coverage, no written conventions, a .env still lingering in the repo. Module after module, you tool up the team with Claude Code.
Free to read
Cheatsheet: 37 slash commands worth knowing
An excerpt from modules 4 and 5, which cover every built-in command with its aliases, its arguments and the right moment to type it.
Session and navigation
/clear [name]New conversation with an empty context · aliases /reset, /new/resume [session]Resume a conversation by ID or name · alias /continue/rename [name]Rename the current session/branch [name]Branch the conversation here; the original stays reachable/rewindRewind the code and/or the conversation · aliases /checkpoint, /undo/export [file]Export the conversation as plain text
Context and costs
/context [all]Context usage, saturation warnings, suggestions/compact [instructions]Summarize the conversation, focusing on what to keep/autocompact [auto|tokens]Automatic compaction threshold, saved in settings/btw [question]Side question that stays out of the history/usageSession cost, plan limits, statistics · aliases /cost, /stats
Model and effort
/model [model]Switch model and save it as the default/effort [level]low, medium, high, xhigh, max, ultracode or auto/fast [on|off]Toggle fast mode/advisor [model|off]Enable an advisor model consulted at key moments
Project and memory
/initGenerate a starter CLAUDE.md/memoryEdit CLAUDE.md and CLAUDE.local.md for each scope/add-dir <path>Add a working directory to the session/cd <path>Change directory without losing the conversation
Code quality
/diffWorking-tree changes, Claude ’s edits included/code-review [level] [--fix]Code review, with optional fixes/security-reviewSecurity analysis of the branch: injections, auth, exposed data/simplify [target]Four parallel agents simplify the code/verifyBuilds, runs and drives the app to verify a change
Plan and parallelism
/plan [description]Switch to plan mode immediately/fork [prompt]Copy the conversation to a background session/subtask <task>Forked subagent that inherits the conversation/batch <instruction>Split a large change across parallel agents/tasksBackground work in this session · alias /bashes/loop [interval] [prompt]Repeat a prompt at a regular interval
Security and tooling
/permissionsAllow / ask / deny rules per scope · alias /allowed-tools/sandboxToggle sandbox mode/hooksHooks on tool events/skillsList available skills; /skill-doctor for their cost/mcpMCP server connections/doctorDiagnostics: installs, PATH, settings, slow hooks · alias /checkup/install-github-appInstall the GitHub app, workflows and secrets
All hundred commands, grouped by family with aliases, arguments and minimum versions, are in modules 4 and 5.
Open module 4Syllabus
Sixteen modules, one project, one exam
The overview and the first two modules are free to read. The rest unlocks with a Premium plan.
- 0Course overview, prerequisites and the Kiosque running exampleFree
- 1Install Claude Code and nail your first sessionFree
- 2The agentic loop, built-in tools and the context windowFree
- 3CLAUDE.md, rules and memory: teach the project to ClaudePremium
- 4Built-in slash commands (1/2): session, context, configuration and modelPremium
- 5Built-in slash commands (2/2): quality, parallelism, automation and integrationsPremium
- 6Build your own slash commands: skills from A to ZPremium
- 7Permissions, execution modes, sandbox and settings filesPremium
- 8Plan mode, checkpoints, sessions and worktrees: working without fearPremium
- 9Hooks: automate and lock down Claude’s behaviorPremium
- 10Subagents, parallel agents and agent teamsPremium
- 11MCP: connect Claude Code to your tools and your dataPremium
- 12Plugins and marketplaces: package and share your toolingPremium
- 13Daily workflows: explore, fix, test, review, shipPremium
- 14Headless mode, GitHub Actions CI and the Agent SDKPremium
- 15Master costs, security and team deploymentPremium
- 16Project: the Kiosque team’s complete Claude Code toolkitPremium
- 17Recap and 40-question final examPremium
Frequently asked questions
Before you start
To do the exercises, yes: a Claude account (Pro, Max, Team or Enterprise) or an Anthropic API key. Reading the course and the cheatsheet requires nothing.
The content follows the official Claude Code documentation as of September 2026, when Claude Fable 5.1 shipped. Every command, every frontmatter key and every hook event we cite really exists; when a behavior depends on the version, the course says so.
About 8 hours: sixteen modules, a running project and a 40-question final exam.
Being comfortable in a terminal and with Git, and able to read Python and TypeScript. The running example uses FastAPI and React without requiring expertise in either framework.
Yes. The overview and modules 1 and 2 are free to read, as is the cheatsheet on this page. The fourteen other modules and the exam are reserved for Premium plans.
Yes. When you pass the exam, a certificate of completion is issued immediately; its number is verifiable on the platform.
No. InSkillML is an independent training provider. Claude and Claude Code are trademarks of Anthropic, PBC; they are cited to describe the content being taught.
Start with module 1. It’s free.
Install Claude Code, read what a session costs, nail your first agentic loop. The rest of the course is waiting.
Independent course, not affiliated with Anthropic. Claude and Claude Code are trademarks of Anthropic, PBC.