M1 · Foundations
What Claude Code is, how the agent loop works, and where the five extension mechanisms sit. Sets the shared vocabulary before any code is written.
git clone https://github.com/huangjia2019/claude-code-engineering.git
AI-Assisted Programming Workshop
A modular workshop curriculum I have run for keysight, LSEG, QECon Shenzhen, the Singularity Intelligence Conference, Packt and Geekbang. The material is organized along two axes: the five extension mechanisms of Claude Code on one side, seven self-contained teaching modules on the other. Everything is in one public repository — fork it, strip what you do not need, run the workshop your way.
The workshop is built on a small cross product. The horizontal axis lists the five extension mechanisms taught in the Claude Code book; the vertical axis lists the seven teaching modules. Each module exercises one or more mechanisms against a real engineering scenario.
Sub-Agents Skills Slash Cmds Hooks MCP
M1 Foundations . . . . .
M2 Sub-Agents X . . . .
M3 Skills . X . . .
M4 Slash Commands . . X . .
M5 Hooks . . . X .
M6 MCP Integration . . . . X
M7 Team Playbook X X X X X
Each row is a half-day module. The last module composes all five mechanisms into a team-level playbook.
Seven modules, each runnable on its own. Click through to the module folder on GitHub for slides, lab scripts, and the teacher’s notes.
What Claude Code is, how the agent loop works, and where the five extension mechanisms sit. Sets the shared vocabulary before any code is written.
git clone https://github.com/huangjia2019/claude-code-engineering.git
Task delegation with isolated context. Building a code reviewer and a test runner that survive a real PR pipeline.
cd workshop/modules/M2-sub-agents && cat README.md
Declarative capability packages defined through SKILL.md. Writing a financial-analysis skill and an API-generator skill from scratch.
cd workshop/modules/M3-skills && cat README.md
User-invoked workflows. Wrapping common engineering rituals (code review, release notes, on-call handover) into one-line commands.
cd workshop/modules/M4-slash-commands && cat README.md
Event-driven automation around tool execution. Building pre-commit checks, post-edit linters, and a log-analysis hook that runs without user input.
cd workshop/modules/M5-hooks && cat README.md
Model Context Protocol for external tools. Writing a custom MCP server in TypeScript and wiring it into the team’s daily workflow.
cd workshop/modules/M6-mcp && cat README.md
Composing all five mechanisms into a single team workflow: shared skills repository, plugin packaging, governance, and rollout checklist.
cd workshop/modules/M7-team-playbook && cat README.md
Six representative runs of the workshop in 2025 and 2026. Each instance folder contains the slide deck, the lab agenda, anonymized feedback notes, and the customizations made for that audience.
Two-day onsite for a measurement-instrument R&D team. Focus on test generation and instrument-driver scaffolding.
Refinitiv quant and risk teams. Regulatory document analysis, multi-agent ticket review, and Copilot-driven spec development.
Quality engineering conference workshop. Code review, test generation, incident diagnosis and engineering-metrics agents.
Keynote-adjacent workshop on harness engineering and the agent design-pattern catalogue.
Companion workshop for the Packt Claude Code engineering title. Walks the book’s case studies as live labs.
Online column workshop. Compressed three-hour version of M1, M2 and M3, tailored for a self-paced audience.
The workshop is a public repository. To run it yourself, clone the repo and read the process notes — they cover audience sizing, time allocation per module, lab environment, and the customizations that have worked in different settings.
git clone https://github.com/huangjia2019/claude-code-engineering.git
The full fork-and-deliver guide lives at workshop/PROCESS.md. For a private engagement or in-house delivery, the easiest contact path is via the about page.