English | 日本語 | Español

ai-agent-camp

AI Agent Training for Non-Engineers - Complete Guide to Claude Code / Cursor / Codex

⚠️ Before you clone

Official repository (maintained by the authors):

• https://github.com/minicoohei/ai-agent-camp

Running AI agents from this repo grants them shell, file-write, and external-API permissions on your machine. Forks and mirrors can silently alter lessons or skills. Before running any lesson, verify your clone:

python3 tools/scripts/verify_integrity.py

If the check reports a fork or missing files, review git diff upstream/main -- .claude/ skills/ tools/ scripts/ before proceeding. See docs/security-guardrails.md.

Releases & Downloads

Releases are cut from main as tags matching v* (semver). A single tag produces three language-specific zip archives, all attached to the same GitHub Release.

Download URL pattern (pinned, never moves once published):

https://github.com/minicoohei/ai-agent-camp/releases/download/{tag}/ai-agent-camp-{lang}-{tag}.zip

Language	Asset	Checksum
日本語	ai-agent-camp-ja-{tag}.zip	ai-agent-camp-ja-{tag}.zip.sha256
English	ai-agent-camp-en-{tag}.zip	ai-agent-camp-en-{tag}.zip.sha256
Español	ai-agent-camp-es-{tag}.zip	ai-agent-camp-es-{tag}.zip.sha256

Each zip contains courses/, skills/, .claude/, .cursor/, docs/ with language suffixes already stripped, plus a CHECKSUMS.txt covering every file inside.

For humans

# Latest release, any language
gh release download --repo minicoohei/ai-agent-camp --pattern 'ai-agent-camp-en-*.zip'

# Specific version
gh release download v0.1.0 --repo minicoohei/ai-agent-camp \
  --pattern 'ai-agent-camp-en-v0.1.0.zip'

For AI agents

Parse the <!-- AGENT-META v1 --> block above, then:

1. (optional) hit latest_tag_api to resolve the current tag
2. fill release_asset_pattern with {tag} and {lang}
3. download, verify against the matching .sha256 asset, unzip

If you only need the lesson manifest, use manifest_raw_pattern with the lang_suffix map to fetch courses/lessons.manifest[.en|.es].yaml directly from a ref (branch, tag, or commit SHA).

See docs/release-process.md for the full specification, versioning policy, rollback procedure, and a Python parsing sample.

Before the first tag is cut: there are no Release assets yet, so the URL pattern above will 404 until v0.1.0 ships. Until then, clone the repo directly (git clone https://github.com/minicoohei/ai-agent-camp.git) or fetch individual files via raw.githubusercontent.com/.../main/....

Table of Contents

• Project Overview
• Key Features
• Quick Start
• Web Course (Recommended)
• Tool Comparison
• Learning Path
• Directory Structure
• Skill Matrix
• Required APIs
• Releases & Downloads
• Documentation
• FAQ
• Contributing
• Support

---

Project Overview

ai-agent-camp is a comprehensive training resource for non-engineering roles such as marketing, sales, planning, and administration to automate and streamline their work using AI agents like Claude Code, Cursor, and Codex.

Target Audience

• People with no programming experience
• Those who want to use AI tools in their daily work
• Anyone interested in automation and data analysis
• Teams looking to improve their overall AI literacy

Vision

Democratize AI agents from "tools for specialists" to "tools everyone can use," boosting productivity across the entire organization.

---

Key Features

• Optimized for Non-Engineers

  • No programming knowledge required
  • Step-by-step tutorials
  • Materials based on real business scenarios

• Practical Skill Set

  • Ready-to-use commands documented in the Commands Reference
  • Implemented skills documented in the Skills Reference
  • Business-specific workflow samples

• Comprehensive Curriculum

  • Official web catalog at ai-agent.camp (as of 2026-08-31): 47 modules, FAQ study time about 50–60 hours
  • This repository: 26 Foundation chapters, 6 Setup chapters, 26 core modules (numbered 1–25 and 29; 26–28 unused), 4 practice chapters — 62 course chapters
  • 138 runnable start-* lessons in courses/lessons.manifest.yaml
  • course.yaml estimates 50 hours; the public FAQ uses 50–60 hours for the full web catalog

• Security and Best Practices

  • API key management guidelines
  • Secure data handling methods
  • Enterprise deployment guidance

• Business Workflow Templates

  • Customer support
  • Sales processes
  • Content marketing
  • Onboarding automation
  • Approval flow optimization

---

Quick Start

Prerequisites

• Git installed
• Python 3.9+
• Internet connection
• Cursor, Claude Code, or Codex available
• Windows and macOS are both supported. Some features (MCP integrations, for example) can differ by OS. For CLI-heavy workflows in this repository, docs/terminal-guide.md documents WSL2 + Ubuntu as a recommended Windows path.

Which tool should I choose? Run /start-0-8 (Tool Selection Guide) to compare Cursor / Claude Code / Codex and find the best fit.

Quickest Way to Begin

1. Clone this repo
2. Read the entry point for your tool:
   • Codex: AGENTS.md
   • Claude Code: CLAUDE.md
   • Cursor: .cursor/commands/lesson/start-0-1.md and other files in .cursor/commands/*
3. Review the safety rules:
   • Codex: docs/codex-safety.md
   • Claude Code / Cursor: docs/security-guardrails.md
4. Run setup verification:
   • Codex: aiagent-check-setup
   • Cursor: /check-setup
5. Start the first lesson with start-0-1

Create Your Own Repository

Copy this repository as your own private repo using one of these methods:

Method 1: Import Repository (GUI - Easy)

This can be done entirely through the GitHub web interface.

1. Log in to GitHub and select "+" > "Import repository" in the top right
2. Enter:
   • Your old repository's clone URL: https://github.com/minicoohei/ai-agent-camp.git
   • Repository name: any name (e.g., my-aiagent)
   • Privacy: Select Private
3. Click "Begin import"
4. After import completes, clone your repository:

   git clone https://github.com/{your-username}/my-aiagent.git
   cd my-aiagent
   

Method 2: Clone & Push (Command Line)

For those comfortable with the terminal.

# 1. Create an empty Private repository on GitHub

# 2. Create a mirror clone
git clone --bare https://github.com/minicoohei/ai-agent-camp.git my-aiagent.git
cd my-aiagent.git

# 3. Set the new origin and push
git push --mirror https://github.com/{your-username}/my-aiagent.git

# 4. Clone again for regular use
cd ..
git clone https://github.com/{your-username}/my-aiagent.git

Pulling Updates from the Original Repository

When the source materials are updated, pull changes with:

# First time only: add the original repo as upstream
git remote add upstream https://github.com/minicoohei/ai-agent-camp.git

# Pull updates
git fetch upstream
git merge upstream/main

If you use Cursor, you can run /update-material in chat to do the same thing.

Note: If you have made local changes, merge conflicts may occur. Resolve them manually.

Installation Steps

1. Clone the Repository

git clone https://github.com/minicoohei/ai-agent-camp.git ~/ai-agent-camp
cd ~/ai-agent-camp

2. Set Up Environment Variables

# First, prepare the key entry in .env.local
uv run python tools/credential_manager.py prepare-dotenv GEMINI_API_KEY

# After saving, optionally migrate to the Credential Store
# uv run python tools/credential_manager.py import-dotenv GEMINI_API_KEY --delete

Do not paste API keys in chat. Save them in .env.local.

3. Install Python Dependencies

# Install dependencies with uv
uv sync

4. Start the Course

# Open the course in a browser
# macOS: open https://ai-agent.camp
# WSL2:  wslview https://ai-agent.camp   (or just open the URL in your Windows browser)

# Open the workspace in Cursor
cursor .

# Claude Code and Codex also work with the same repo
claude
codex

Getting Started by Tool

Cursor

/check-setup
/overview
/start-0-1

Claude Code

• Read CLAUDE.md
• Review docs/security-guardrails.md
• After setup verification, proceed to the lesson flow

Codex

• Read AGENTS.md
• Open docs/codex-guide.md
• Run aiagent-check-setup skill for environment verification
• Pass start-0-1 to the aiagent-lesson-runner skill to start the first lesson

---

Web Course (Recommended)

Want a guided, structured learning experience?

AI Agent Camp is the official web course. As of 2026-08-31 the public catalog lists 47 modules, and the FAQ estimates about 50–60 hours to finish. A 24/7 AI tutor and a desktop app for environment setup are available on the site.

This repository is the Cursor / Claude Code / Codex workspace for the same curriculum. It ships 138 start-* lessons (see courses/lessons.manifest.yaml). Copilot / ACP is not a first-class track here.

👉 Start learning at ai-agent.camp

---

Tool Comparison

The materials themselves are shared. The differences are in how you enter and interact.

Item	Codex	Claude Code	Cursor
Entry point	AGENTS.md	CLAUDE.md	.cursor/commands/*
Start lesson	aiagent-lesson-runner	Claude lesson flow	/start-*
Setup check	aiagent-check-setup	Claude-side check	/check-setup
Safety model	sandbox + approval	Claude hooks + permissions	Cursor rules + commands
Lesson ID	start-*	start-*	start-*

Common points:

• All tools use the same repository
• All tools use the same lesson IDs
• Security principles for secrets and Git are the same

Basic rules for learners:

• Keep small tasks lightweight
• Write a short plan before starting large tasks
• Read the relevant safety guide before touching Git, MCP, or secrets

---

Learning Path

Phase 1: Foundation (AI Fundamentals) - 5 hours

Learn the basics of AI agents. The path below covers the first 11 chapters (about 5 hours); the Foundation lesson contains 26 chapters in total.

Chapter	Content	Study Time
0-1	How LLMs Work - Fundamental Principles	30 min
0-2	Token Concepts and Calculation	30 min
0-3	What Are AI Agents	30 min
0-4	Context Engineering and Prompting	30 min
0-5	How to Use Cursor	25 min
0-6	MCP (Model Context Protocol)	25 min
0-7	Multimodal AI	25 min
0-8	RAG (Retrieval-Augmented Generation)	25 min
0-9	Skills / SubAgents / Agent Teams	25 min
0-10	Hallucination Prevention	25 min
0-11	AI Security	25 min

Learning Objectives

• Understand the fundamentals of LLMs
• Know what tokens are
• Grasp AI agent concepts
• Write better prompts
• Understand how MCP, RAG, and multimodal systems work
• Recognize security risks when using AI

---

Phase 2: Setup (Environment Setup) - 1.5 hours

Prepare to actually use AI agents.

Module	Content	Study Time
0	Claude Code / Cursor Setup	30 min
0.5	Extensions and Customization	15 min
0.9	API Key Setup and Authentication	45 min

Learning Objectives

• Install Claude Code / Cursor
• Manage API keys securely
• Complete basic configuration

---

Phase 3: Core Modules

Acquire skills you can use in real work. Lesson counts below match courses/lessons.manifest.yaml (modules 26–28 are unused).

#	Module	Key Skills	Lessons	Difficulty
1	Banner & Image Generation	banner-creator, nanobanana	3	★
2	Diagrams & Flowcharts	diagram-generator, PlantUML	3	★★
3	Tutorials	screenshot-analyzer, tutorial-generator	6	★★
4	Google Workspace	gogcli, Gmail, Calendar, Drive, Sheets	7	★★★
5	PPTX Analysis & Editing	pptx-analyzer, pptx-creator, pptx-converter	2	★★
6	Agent Development	Commands/Skills creation, customization	9	★★★★
7	Skills & Commands	Skill design, SKILL.md implementation, testing, design patterns	8	★★★★
8	Data Analysis & EDA	data-analyst, BigQuery, Marimo	4	★★★
9	Slack Integration	slack-search, check-inbox, task-manager	3	★
10	GAS Automation	gas-clasp-ops, Calendar, Sheets	4	★★★
11	GitHub Actions	Workflow, Secrets, CI/CD	5	★★★
12	Notion Integration	Notion MCP, DB operations, ncli	6	★★
13	LP/HP Creation	Messaging, wireframes, Pencil design, HTML, Vercel deploy	5	★★★
14	Article Writing	article-writer, copy-editing, fact-checker	7	★★★
15	Video Production	Kling, HeyGen, Veo, Remotion, MV	13	★★★
16	Email/LINE Automation	email-sequence, Resend, LINE API	8	★★★
17	Marketing	X posts, SEO research, copywriting, design mocks	5	★★★
18	Requirements & System Dev	pm-toolkit, test-planner, Notion integration	20	★★★★★
19	Microsoft Office (Outlook)	Outlook MCP integration	2	★★
20	Freee/MoneyForward	Freee MCP accounting data operations	1	★★
21	Figma	Figma MCP	1	★★
22	Discord	Discord Bot / channel operations	1	★★
23	LINE	LINE official account	1	★★
24	Salesforce	Salesforce CLI	1	★★
25	Google Ads	Google Ads API	1	★★
29	slide-forge	Slide generation and revision	4	★★

This repository: 138 start-* lessons (including Module 0 setup). Web course catalog (ai-agent.camp, as of 2026-08-31): 47 modules; FAQ study time about 50–60 hours.

Learning Options

Recommended order (Beginners)

Module 1 → Module 2 → Module 3 → Module 5 → Module 6 → Module 8

Recommended order (Business Efficiency Focus)

Module 4 → Module 9 → Module 10 → Module 11 → Module 12 → Module 8

Recommended order (Creative Focus)

Module 1 → Module 2 → Module 3 → Module 15 → Module 13 → Module 14

Recommended order (Marketing Focus)

Module 1 → Module 17 → Module 13 → Module 15 → Module 14 → Module 16

---

CursorBootcamp YAML Metadata

YAML metadata for the CursorBootcamp platform is stored in courses/aiagent/. Some chapters include practice/ (exercises) and final/ (final assignments); chapters that link out to the web materials do not.

Lesson	Chapters	Content
Lesson 01: Foundation	26	LLM basics, Tokens, Agents, Context Engineering, MCP, Multimodal, RAG, SubAgents, Hallucination, Security, Observability, Feasibility
Lesson 02: Setup	6	Environment setup, Extensions, API configuration, Security check, Remote environment, Knowledge base
Lesson 03: Core	26	Banner through slide-forge (all core modules)
Lesson 04: Practice	4	Diagram drill, parallel sessions, research-to-proposal deck, PMO operations (no practice/ / final/ directories)

62 chapters in total.

---

Directory Structure

ai-agent-camp/
│
├── courses/                              # Curriculum source of truth
│   ├── lessons.manifest.yaml             # Lesson manifest
│   └── aiagent/
│       ├── course.yaml / .en.yaml / .es.yaml  # Course definition (multilingual)
│       ├── cover.png                     # Course cover image
│       ├── lesson01-foundation/          # Fundamentals (26 chapters)
│       │   └── ch00 ~ ch24/             # LLM, Tokens, Agents, MCP, RAG, Security, Observability, Feasibility, etc.
│       ├── lesson02-setup/              # Environment setup (6 chapters)
│       │   └── ch01 ~ ch06/             # Environment, Extensions, API setup, Security check, Remote env, Knowledge base
│       ├── lesson03-core/               # Core skills (26 available modules)
│       │   └── module01 ~ module25, module29/ # Banner through slide-forge
│       └── lesson04-practice/           # Practice exercises (4 chapters)
│           └── ex01 ~ ex04/             # Diagram drill, parallel sessions, research deck, PMO ops
│
├── .cursor/commands/                     # See docs/commands-reference.md
│   ├── lesson/                          # Learning commands
│   │   ├── /start-0-1 ~ /start-0-8     # Module 0: Setup
│   │   ├── /start-1-1 ~ /start-1-3     # Module 1: Banners
│   │   ├── /start-2-1 ~ /start-2-3     # Module 2: Diagrams
│   │   ├── /start-3-1 ~ /start-3-6     # Module 3: Tutorials
│   │   ├── /start-4-1 ~ /start-4-7     # Module 4: Google Workspace
│   │   ├── /start-5-1 ~ /start-5-2     # Module 5: PPTX
│   │   ├── /start-6-1 ~ /start-6-9     # Module 6: Agent Development
│   │   ├── /start-7-1 ~ /start-7-8     # Module 7: Skills/Commands
│   │   ├── /start-8-1 ~ /start-8-4     # Module 8: Data Analysis
│   │   ├── /start-9-1 ~ /start-9-3     # Module 9: Slack
│   │   ├── /start-10-1 ~ /start-10-4   # Module 10: GAS
│   │   ├── /start-11-1 ~ /start-11-5   # Module 11: GitHub Actions
│   │   ├── /start-12-1 ~ /start-12-6   # Module 12: Notion
│   │   ├── /start-13-1 ~ /start-13-5   # Module 13: LP Creation
│   │   ├── /start-14-1 ~ /start-14-7   # Module 14: Article Writing
│   │   ├── /start-15-1 ~ /start-15-13  # Module 15: Video Production (+ 7a-7d)
│   │   ├── /start-16-1 ~ /start-16-8   # Module 16: Email/LINE Automation
│   │   ├── /start-17-1 ~ /start-17-5   # Module 17: Marketing
│   │   ├── /start-18-1 ~ /start-18-20  # Module 18: Requirements/System Dev
│   │   ├── /setup-m365cli + /start-19-1 ~ /start-19-2 # Module 19: Outlook
│   │   ├── /setup-freee + /start-20-1   # Module 20: Freee/MoneyForward
│   │   ├── /setup-figma + /start-21-1   # Module 21: Figma
│   │   ├── /setup-discord + /start-22-1 # Module 22: Discord
│   │   ├── /setup-line-harness + /start-23-1 # Module 23: LINE
│   │   ├── /setup-salesforce + /start-24-1 # Module 24: Salesforce
│   │   ├── /setup-google-ads + /start-25-1 # Module 25: Google Ads
│   │   └── /start-29-1 ~ /start-29-4   # Module 29: slide-forge
│   │
│   └── utility/                         # Utility commands
│       ├── /check-setup                 # Setup verification
│       ├── /overview                    # Project overview
│       ├── /guide                       # Usage guide
│       ├── /tutor                       # Interactive help
│       ├── /update-material             # Update materials to latest
│       └── ... other helpers
│
├── skills/                              # See docs/skills-reference.md
│   │
│   │  -- Image & Banner Generation --
│   ├── banner-creator/                  # SNS banner generation
│   ├── nanobanana/                      # General image generation/editing
│   ├── diagram-generator/              # Infographic generation
│   │
│   │  -- Screenshots & Tutorials --
│   ├── screenshot-analyzer/             # Screenshot analysis
│   ├── screenshot-annotator/            # Screenshot annotation
│   ├── tutorial-generator/              # Auto tutorial generation
│   │
│   │  -- Document Processing --
│   ├── pptx-analyzer/                   # PPTX structure analysis
│   ├── pptx-converter/                  # PPTX template conversion
│   ├── pptx-creator/                    # Topic to PPTX auto-generation
│   ├── document-processor/              # PDF/Word processing
│   ├── pdf-compressor/                  # PDF compression
│   │
│   │  -- Video & Media --
│   ├── storyboard-generator/            # Storyboard + Kling video generation
│   ├── video-frame-reader/              # Video keyframe extraction
│   ├── media-generator/                 # Media file generation
│   │
│   │  -- Data Analysis & Auth --
│   ├── data-analyst/                    # Data analysis & EDA
│   ├── bigquery-auth/                   # BigQuery authentication
│   ├── gcp-auth/                        # GCP authentication & setup
│   │
│   │  -- Slack & Communication --
│   ├── check-inbox/                     # Email/Slack TODO extraction
│   ├── slack-search/                    # Slack semantic search
│   ├── slack-task-manager/              # Slack task management
│   ├── slack-unanswered/                # Unanswered message detection
│   │
│   │  -- GAS & Others --
│   ├── gas-clasp-ops/                   # Google Apps Script operations
│   ├── lp-designer/                     # LP/HP creation workflow
│   │
│   │  -- Marketing (20 skills) --
│   ├── ab-test-setup/                   # A/B test design & implementation
│   ├── analytics-tracking/              # GA4/GTM tracking
│   ├── competitor-alternatives/          # Competitor comparison pages
│   ├── content-strategy/                # Content strategy
│   ├── copy-editing/                    # Copy editing & review
│   ├── copywriting/                     # Marketing copy
│   ├── email-sequence/                  # Email sequences
│   ├── free-tool-strategy/              # Free tool strategy
│   ├── launch-strategy/                 # Launch strategy
│   ├── marketing-ideas/                 # Marketing ideas
│   ├── marketing-psychology/            # Marketing psychology
│   ├── paid-ads/                        # Paid ad campaigns
│   ├── pricing-strategy/                # Pricing strategy
│   ├── product-marketing-context/       # Product marketing
│   ├── programmatic-seo/                # Programmatic SEO
│   ├── referral-program/                # Referral programs
│   ├── schema-markup/                   # Structured data
│   ├── seo-audit/                       # SEO audits
│   └── social-content/                  # Social media content
│
├── mv-composer/                         # Operations/production-only Remotion video tools (not required for learners)
├── gas-example/                         # Operations/production-only GAS examples (not required for learners)
├── ops/                                 # Operations/production-o