Name: Claude Code: Above and Beyond
Price: 127.00 USD
Availability: InStock

About Course

Claude Code: Above and Beyond

Go deeper into agentic development and build the setup Claude Code works in

Course Summary

Claude Code: Essentials taught you to work with the agent. This course teaches you to build the setup it works in, so good results stop depending on how carefully you prompt. You turn your team’s rules into hooks and permissions the agent cannot skip, hand repeatable work to skills and subagents, connect it to your tools with MCP, and run it in CI with nobody watching.

CLAUDE.md can tell the agent never to hand-edit ModelSnapshot.cs, and in a long task it edits it anyway. A hook blocks that edit every time.

What Will You Learn?

Turn the rules that matter into hooks and permissions that enforce them
Write skills for your team’s repeatable .NET tasks
Use subagents to review diffs and write tests
Run parallel work in worktrees and merge it cleanly
Connect MCP servers, and build your own in C#
Run Claude Code safely in CI, with hard limits on what it can do
Package your setup as a plugin and roll it out to your team

Course Description

The course builds on the same Repair Desk solution as Essentials. Every module ends with two versions of its last step, one on Repair Desk and one on your own codebase, so you finish with a working setup for a real .NET project.

It is the second of the two parts. Start with Claude Code: Essentials, or come with the same habits from daily use.

You can find the full plan of all 78 lessons in the Course Content section below.

Dev, the little red devil with horns and a trident, and Angy, the white angel with a halo and wings, each wave with their outer hand on either side of a tilted yellow sign that reads WELCOME

Course Content

Module 0: Course Introduction

What This Course Builds See the finished harness block an edit and review a change in a .NET repository it has never seen, and learn what the course leaves out on purpose.
What You Need From Essentials Pass the five-item self-check, or close the gaps it finds, and set up the final checkpoint of Claude Code: Essentials as your starting point.
Course Map and Fast Path Plan your route: every module with its estimated hours, or the 23-lesson fast path that ends in a packaged harness.
Repair Desk or Your Own Solution Choose between Repair Desk and your own .NET solution, and get your own solution ready if you pick it.
Setup, Versions and Conventions Get WSL 2 ready for the sandboxing module, compare your Claude Code version with the course's, and learn the lesson conventions that are new here.

Module 1: From Prompting to Environment Design

Module 2: Writing and Measuring Skills

What a Skill Is Explain what a SKILL.md file is, where skills live, and how one runs: as your slash command or when Claude picks it.
The /add-endpoint Skill Move your Minimal API conventions out of CLAUDE.md into a skill, and add an endpoint with it.
Skill Arguments and Dynamic Context Pass arguments to a skill, and feed it live output such as git diff and dotnet --version before it runs.
The /add-migration Skill Build the safe EF Core migration steps into a skill, and limit its tools, model and context in the frontmatter.
Skill Invocation Control Decide whether you, Claude or both can start a skill, and write a description that gets it picked at the right moment.
Measuring Skills and Skill Packs Measure what each skill costs in context, try Microsoft's official dotnet/skills and a community kit, and keep what the numbers support.

Module 3: Hooks That Format, Test and Block

Module 4: Session Hooks, Windows and Mods

Hooks on Windows Write hook handlers that work in PowerShell, and make a guard that cannot start block the action.
SessionStart Hooks Start every session with the solution's state already loaded: git status, the SDK version and pending migrations.
UserPromptSubmit and PreCompact Hooks Shape a prompt on its way in, and keep the facts that must survive compaction.
Prompt Hooks Let a small model check an action against a rule, and know when that beats a shell script.
Hooks in Skills, Agents and Policy Declare hooks in skills, agents, plugins and team policy, and decide who may add one.
IDE Hooks in Rider Use Rider's inspections and formatter to clean up the agent's edits, and know why the CLI hooks still do the blocking.
Writing a Mod Rebuild the database-drop guard as a mod with Proceed and Cancel buttons, and validate and test it before it loads.

Module 5: Subagents for Review and Testing

Module 6: Worktrees, Batches and Workflows

Module 7: Connecting MCP Servers

Module 8: Building an MCP Server in C#

Module 9: Permissions, Sandboxing and Security

Module 10: Headless Mode and CI

Module 11: Goals, Loops and Routines

Module 12: Plugins and Team Rollout

Module 13: Final Project and Next Steps

Student Ratings & Reviews

No Review Yet

About Course

Claude Code: Above and Beyond

Course Summary

What Will You Learn?

Course Description

Course Content

Module 0: Course Introduction

What This Course Builds See the finished harness block an edit and review a change in a .NET repository it has never seen, and learn what the course leaves out on purpose.

What You Need From Essentials Pass the five-item self-check, or close the gaps it finds, and set up the final checkpoint of Claude Code: Essentials as your starting point.

Course Map and Fast Path Plan your route: every module with its estimated hours, or the 23-lesson fast path that ends in a packaged harness.

Repair Desk or Your Own Solution Choose between Repair Desk and your own .NET solution, and get your own solution ready if you pick it.

Setup, Versions and Conventions Get WSL 2 ready for the sandboxing module, compare your Claude Code version with the course's, and learn the lesson conventions that are new here.

Module 1: From Prompting to Environment Design

Advisory Versus Deterministic Rules Watch a CLAUDE.md rule fail under pressure, then block the same edit with a ten-line hook.

Rules, Skills, Hooks and Subagents Tell the four instruments apart, see where auto memory fits, and pick the right one with a decision table.

The .claude Directory Know what goes in the .claude directory, what to commit, what stays on your machine, and where auto memory lives instead.

A Harness Plan for Repair Desk Match each known Repair Desk failure to the instrument that stops it, and save the plan in the repository.

Scaling the Harness to Large Solutions Adapt the harness to a solution of forty projects: rules per area, one project per build, and filtered test runs.

Module 2: Writing and Measuring Skills

What a Skill Is Explain what a SKILL.md file is, where skills live, and how one runs: as your slash command or when Claude picks it.

The /add-endpoint Skill Move your Minimal API conventions out of CLAUDE.md into a skill, and add an endpoint with it.

Skill Arguments and Dynamic Context Pass arguments to a skill, and feed it live output such as git diff and dotnet --version before it runs.

The /add-migration Skill Build the safe EF Core migration steps into a skill, and limit its tools, model and context in the frontmatter.

Skill Invocation Control Decide whether you, Claude or both can start a skill, and write a description that gets it picked at the right moment.

Measuring Skills and Skill Packs Measure what each skill costs in context, try Microsoft's official dotnet/skills and a community kit, and keep what the numbers support.

Module 3: Hooks That Format, Test and Block

Hook Events and Matchers Choose the hook event for a job, and match it to the tools you want it to watch.

Exit Codes and Hook Testing Predict what each exit code does, and test a hook before you trust it, on Windows too.

Formatting After Every Edit Run dotnet format on every file the agent edits, with .editorconfig as the source of truth.

Tests Before the Agent Stops Keep the agent working until dotnet test passes, without letting a failing suite loop forever.

Blocking Dangerous Edits Block edits to generated EF Core files, hand-edited package versions and database drops, with a reason the agent can read.

Hooks and Permission Rules Predict which one wins when a hook and a permission rule disagree, and see what is loaded with /hooks.

Module 4: Session Hooks, Windows and Mods

Hooks on Windows Write hook handlers that work in PowerShell, and make a guard that cannot start block the action.

SessionStart Hooks Start every session with the solution's state already loaded: git status, the SDK version and pending migrations.

UserPromptSubmit and PreCompact Hooks Shape a prompt on its way in, and keep the facts that must survive compaction.

Prompt Hooks Let a small model check an action against a rule, and know when that beats a shell script.

Hooks in Skills, Agents and Policy Declare hooks in skills, agents, plugins and team policy, and decide who may add one.

IDE Hooks in Rider Use Rider's inspections and formatter to clean up the agent's edits, and know why the CLI hooks still do the blocking.

Writing a Mod Rebuild the database-drop guard as a mod with Proceed and Cancel buttons, and validate and test it before it loads.

Module 5: Subagents for Review and Testing

Subagent Context Isolation Explain what a subagent keeps out of your session, what comes back, and what it costs.

The Explore and Plan Agents Recognise when Claude hands work to its built-in Explore and Plan agents, and read what they return.

The Diff Reviewer Agent Write a reviewer agent that sees only the diff and the acceptance criteria, reports gaps in the change and leaves style alone.

Cheaper Models for Test Writing Run a test-writing agent on a cheaper model with a limited tool set and a turn limit.

Adversarial Review of a Data Migration Catch an unsafe shortcut in a data migration with a reviewer agent, and write the criteria it was missing.

Background Subagents and /subtask Run subagents in the background, use /subtask, and spot the cases where delegating only costs tokens.

Module 6: Worktrees, Batches and Workflows

Merging Work From Two Worktrees Merge two agent-written branches in the right order when both touch a migration or the packages file.

Measuring Your Review Capacity Time your own review of two worktrees, and plan parallel work around the number of diffs you can really read.

The /batch Command Apply one mechanical change across many units, each in its own worktree and pull request, after testing it on two.

Dynamic Workflows Start a workflow run, watch its agents, pause or stop it, and check the result against your hooks.

Choosing a Parallel Tool Choose between subagents, worktrees, batches, workflows and agent teams by the shape of the task.

Module 7: Connecting MCP Servers

The Microsoft Learn MCP Server Have the agent answer a .NET 11 question from the current Microsoft Learn documentation.

What MCP Is Explain MCP servers, tools, resources, prompts and the two transports, and how a tool reaches the model.

Server Scopes and Approval Add a server at local, project or user scope, and know when Claude Code asks you to approve it.

A Read-Only Database Server Connect a read-only server to the SQLite database, and back it with a deny rule.

Aspire and Claude Code See what the Aspire MCP server exposes in a separate Aspire starter, and why Aspire prefers skills for most agent work.

Vetting and Managing MCP Servers Log in to servers, stop them crowding your context, and check a server before you add it.

Module 8: Building an MCP Server in C#

The Official C# MCP SDK Create an MCP server project with the official C# SDK, and register a first tool over stdio.

The get_endpoint_map Tool Return the API's routes, verbs and handlers from its own endpoint data, so the agent stops guessing them.

Finding Callers With Roslyn Find every caller of a method through the compiler, and compare the result and its token cost with grep.

Connecting and Debugging the Server Connect your server to Claude Code, and debug the handshake when no tools show up.

Resources and Prompts From C# Expose your coding conventions as a resource and your review checklist as a prompt.

Packaging the Server for NuGet Pack the server as a .NET tool, version it for your team, and run it with dnx.

Module 9: Permissions, Sandboxing and Security

The Permission Rule Engine Predict how rules handle compound commands, wrappers and Windows paths, and write rules that hold.

Sandboxing on WSL 2 Run Claude Code sandboxed on WSL 2 with filesystem and network limits, and know what native Windows users do instead.

Auto Mode Internals Explain what the auto mode classifier sees and refuses, and when to switch back to Manual.

Defending Against Prompt Injection Recognise injection through a README, a tool result and a web page, and stop each one with a named defence.

Managed Settings and Cloud Providers Set a team policy in managed settings, and answer what changes when the team runs Claude Code through a cloud provider.

Module 10: Headless Mode and CI

Headless Runs With claude -p Run Claude Code from a script with one prompt and one result, and read its output formats and exit codes.

Structured Output and Hard Limits Get output that matches a JSON schema, cap turns and spending, and run with nobody there to approve.

Calling Claude Code From C# Tell the Anthropic NuGet package apart from the agent loop, and drive Claude Code from C# with a small Process wrapper.

Batch Runs Over a File List Apply a change to a list of files from a script, test it on two first, and keep the log as an audit trail.

GitHub Actions for a .NET Repository Run Claude Code in GitHub Actions, with build and test as the checks it has to pass.

Code Review in CI Compare the ways to review every pull request in CI, what each one costs and who pays.