A practical, open-source resource for getting real work done with Claude Code. Built from working projects — not a pitch, not a summary. The docs explain how things work; the skills, templates, and prompts are files you can drop straight into your own setup.
docs/ the course — start here if you're new, jump around if you're not
skills/ 36 installable Claude Code skills — general tools + business automation
templates/ drop-in assets: CLAUDE.md variants, agent definitions, hooks, MCP configs
reference/ fast lookups: command cheatsheet, skills catalog, glossary, troubleshooting
MUST_LISTEN.md curated videos — the ones actually worth your time
General skills (work anywhere, no API keys needed):
| Skill | What it does |
|---|---|
skills/grill-me/ |
Structured interview that extracts knowledge and saves context files |
skills/session-handoff/ |
Pre-/clear snapshot — next session picks up exactly where this left off |
skills/skill-builder/ |
Builds and audits other skills following official best practices |
skills/agent-builder/ |
Builds and audits sub-agent definitions |
skills/excalidraw-diagram/ |
Generates editable Excalidraw diagrams |
skills/frontend-design/ |
Distinctive, production-grade frontend code |
skills/video-to-website/ |
Turns a video into a scroll-driven animated website |
Automation skills (lead gen, cold email, Gmail, YouTube, research, and more):
→ skills/automation/ — 27 skills, grouped by category
Image generation (need a kie.ai API key — cost per image, see each skill's setup):
skills/excalidraw-visuals/ (hand-drawn-style PNGs) and
skills/nano-banana-images/ (hyper-realistic photos).
Full catalog with descriptions and costs: skills/README.md.
| What you need | Where |
|---|---|
| A generic project CLAUDE.md | templates/claude-md/generic.md |
| AIOS / second-brain CLAUDE.md | templates/claude-md/aios.md |
| Trigger.dev project CLAUDE.md | templates/claude-md/trigger-dev.md |
| Sub-agent definitions (code reviewer, researcher, QA, etc.) | templates/agents/ |
| WAT project skeleton | templates/wat-starter/ |
| Full build prompts (GMaps pipeline, Upwork scraper, etc.) | templates/system-prompts/ |
| Step | Page | What you'll get |
|---|---|---|
| 1 | Why Claude Code | What it actually is and why it's different |
| 2 | Install & setup | Plans, install command, login |
| 3 | Where to run it | Terminal, desktop app, VS Code, Antigravity, web |
| 4 | Your first project | Plan mode, permission modes, building something real |
Not ready to pay? → Running Claude Code for free — 18 alternative providers via a local proxy.
Foundations
- What CLAUDE.md is
- Context management — context window, compaction, the 18 token hacks
- Permission modes — plan / ask / auto-edit / bypass
- Tokens & cost
- Plans & usage — Pro vs Max vs Team, model access
- Session management —
--continue,--resume,/rewind
Make it yours
- Slash commands
- Skills — how skills work, progressive loading, building your own
- Hooks
- Subagents
- Agent teams — enable flag, cost reality, when to use
- Persistent memory —
#shortcut, auto-memory, scopes - Rules files —
.claude/rules/, splitting out of CLAUDE.md - Status line
- Settings reference
Connect the world (MCP)
- MCP overview
- Setting up servers
- MCP token cost — what each server actually costs
Advanced
- Git worktrees — parallel sessions without file conflicts
- Parallelization — sub-agent fan-out, failure math, cost calibration
- Multi-agent patterns — assembly line, consensus, debate, verification
- agent-browser — browser automation CLI built for AI agents; deterministic element refs
- Understand Anything — turn any codebase into a knowledge graph before building
- LLM wiki — build a second brain: cross-linked Obsidian wiki ingested from transcripts, PDFs, URLs
- The 5 levels of a second brain — from exact-word CLAUDE.md routing to a fully autonomous knowledge graph
- Plugins & marketplaces
- Headless mode & Agent SDK
- GitHub Actions
Ship it
- Deployment — Modal, Trigger.dev, Vercel, GitHub Actions
- Routines & scheduling — scheduled tasks vs
/loop, memory architecture
Workflows & recipes
- The WAT framework
- Design workflow — screenshot loop, spec-driven design
- Design quality — impeccable + taste-skill; stop generic output at the hook level
- Vibe coding guide — monolith first, lock the backbone, rules file
- Multi-model vibe coding
- Google Workspace CLI — Docs, Gmail, Drive without MCP overhead
- Video editing
- Recipes — social repurposing, multimodal RAG, executive assistant
Business
- Selling AI solutions
- Finding clients — Trojan Horse partner method, outreach
- Pricing workflows — 10x ROI rule, retainer math
- Delivering projects — licensing, security, handover
- Command cheatsheet
- Skills catalog — all 36 skills with env vars
- Agent loop reference
- Glossary
- Troubleshooting
- Resources & links
- Must-watch / must-read
No real API keys anywhere in this repo. Every credential is a placeholder like <YOUR_TOKEN>. Keep your own keys in a local .env — it's in .gitignore already.