agentic-jumpstart

Agentic Jumpstart

A hands-on introduction to building AI workflows with Claude Code. This repo walks you through common concepts and best practices using a simple, guided example — no prior experience required.

This isn't a mandate. It's a place to see patterns in action — skills, hooks, logging, subagents — so you can decide what works for your team and adopt what makes sense.

Who is this for? Everyone. Developers, data engineers, product managers, technical leads — anyone curious about building structured AI workflows. The tutorial is written for all audiences, regardless of technical background.

Get Started

Head to the .starthere/ folder and work through the exercise from the beginning. It should take about 30 minutes.

By the end, you'll have built a working skill, added logging, created a validation hook, and understood how the pieces fit together.

What's Inside

agentic-jumpstart/
├── .starthere/                            # Start here — guided tutorial (~30 min)
├── CLAUDE.md                              # Project-level context (loaded every session)
├── .claude/
│   ├── settings.json                      # Hook configuration (logging, validation)
│   ├── hooks/validators/                  # Validation scripts for hooks
│   ├── skills/meeting-notes-sample/       # Completed example: meeting notes skill
│   ├── skills/sample-skill/               # Template: summarize a folder
│   ├── skills/write-logs/                 # Logging utility (used by all skills/agents)
│   ├── agents/sample-agent.md             # Template: focused subagent
│   └── agents/orchestrator-agent.md       # Template: multi-agent coordinator
├── logs/                                  # Log output directory
├── references/                            # Deeper reference docs
│   ├── building-blocks.md                 # The four primitives of agentic development
│   ├── when-to-use-what.md               # Decision guide: skills vs agents vs hooks
│   ├── logging-guide.md                  # How logging and traceability work
│   └── built-in-variables.md             # Environment variables by component type
├── snippets/
│   └── agentic-snippets.code-snippets    # VS Code snippets for scaffolding components
└── README.md                             # You are here

Reference Docs

Once you've completed the tutorial, these docs go deeper on specific topics:

Document	What You'll Learn
Building Blocks	The four primitives: prompts, agents, tools, and context
When to Use What	Decision guide for choosing between skills, subagents, hooks, and orchestrators
Logging Guide	How the logging system works, how to trace requests, and how to debug with logs
Built-in Variables	Every environment variable available in hooks, skills, and agents

VS Code Snippets

The snippets/ folder contains templates for quickly scaffolding new components:

Prefix	What It Creates
agp	Agentic prompt with full frontmatter and logging
agsk	Skill template with argument handling and built-in variables
agag	Subagent template with skill references and decision logging
agor	Orchestrator template with Task tool spawning pattern
aghk	Hook script with stdin/stdout/stderr handling and exit codes

To install: open VS Code, press Ctrl+Shift+P (or Cmd+Shift+P on Mac), type "Snippets: Configure User Snippets", and paste the contents of snippets/agentic-snippets.code-snippets into a new or existing snippets file.