Claude Resources
Official docs and guides for vibecoders — organized by skill level with TLDRs so you know exactly what to read and when.
Beginner
Just getting started? Start here.
The official "what is this thing" page. Shows Claude's capabilities, use cases, and how it compares. Good first stop.
You want to understand what Claude is before diving in.
The web interface. No setup needed. Just sign up and start chatting. Best way to get a feel for Claude before using the API.
You want to try Claude right now without writing any code.
First API call in 5 minutes. Covers authentication, your first request, and the basic response structure. Assumes zero prior API experience.
You want to call Claude from your own code for the first time.
Lists all current Claude models (Opus, Sonnet, Haiku) with speed/cost/capability tradeoffs. Includes the latest model IDs you need for API calls.
You need to pick the right model for your use case or find the latest model ID.
Free interactive notebooks that walk you through the API fundamentals step by step. Runs in Google Colab — no local setup needed.
You learn best by running actual code and seeing results.
Intermediate
You've got the basics. Now make Claude actually do what you want.
Official guide on how to write prompts that actually work. Covers system prompts, examples, formatting, and common patterns. This is where 80% of your "why isn't it doing what I want" problems get solved.
Claude gives you bad outputs and you want to know why.
Claude Code is an AI coding agent that lives in your terminal. Not just autocomplete — it reads files, edits code, runs tests, and ships. This page explains what it is and what it can do.
You want to understand what Claude Code is and whether it's for you.
How to install and run Claude Code in under 10 minutes. Covers npm install, first run, and the slash commands you'll use daily.
You want to get Claude Code running on your machine right now.
Teach Claude to use tools — web search, database queries, API calls. You define the tool schema, Claude decides when to call it. This is the foundation of any serious Claude app.
You want Claude to do things beyond text — call APIs, query DBs, take actions.
Pass images to Claude and it understands them. Covers supported formats, base64 vs URL, and practical examples like UI screenshots and diagrams.
You need Claude to analyze images, screenshots, or documents.
System prompts set Claude's persona, constraints, and context. This guide covers how to structure them for consistent behavior across conversations.
You're building an app and need Claude to behave consistently every time.
Advanced
Building real products with Claude? Here's what you need.
How to build autonomous Claude agents that plan, act, and loop. Covers the agent loop pattern, tool selection, and stopping conditions. The most important read if you're building anything serious.
You want Claude to complete multi-step tasks autonomously — not just respond once.
MCP is an open standard for connecting Claude to external tools, data sources, and services. Think of it as a plugin system — build once, works with any MCP-compatible AI. This is the future of Claude integrations.
You want Claude to connect to databases, APIs, or local tools in a standardized way.
Hooks let you run shell commands automatically at Claude Code lifecycle events (before/after edits, on commit, etc). Powerful for enforcing workflows, auto-formatting, or blocking bad actions.
You want Claude Code to automatically do things like lint, test, or notify on specific events.
Full reference for settings.json — permission modes, allowed tools, env vars, and project-level vs global config. Read this before granting Claude Code any dangerous permissions.
You want fine-grained control over what Claude Code can and can't do.
Stream Claude's response token-by-token instead of waiting for the full reply. Required for any chat-like UX. Covers SSE format, error handling in streams, and the Python/JS SDK helpers.
You're building a UI that shows Claude's response as it types.
Cache parts of your prompt that don't change (system prompts, large docs) and save up to 90% on token costs. Huge for apps that send the same context every request.
You're burning API credits and the same context is sent on every request.
How to design systems with multiple Claude agents working together — orchestrators, subagents, tool sharing, and communication patterns. Covers parallelization and the tradeoffs of complex agent networks.
You're building a team of agents and need them to coordinate.
Reference
Bookmarks you'll return to constantly.
Full API spec for the Messages endpoint. Every parameter, every response field, every error code. The authoritative source when something doesn't work as expected.
Something is broken and you need to check exact API behavior.
All built-in slash commands (/help, /clear, /compact, /memory, etc) explained. Also covers how to create your own custom slash commands.
You want to know what slash commands exist or how to build your own.
The official Python SDK. Handles auth, retries, streaming, and tool use. Always use this over raw HTTP calls — much safer and maintained.
You're building a Python app with Claude.
Official TS/JS SDK. Full TypeScript types, streaming helpers, and tool use support. Essential for any Node.js or browser-adjacent Claude integration.
You're building a JS/TS app with Claude.
Practical code examples for common Claude use cases — RAG, tool use, vision, structured output, and more. If you want working code for a specific pattern, start here.
You want copy-paste code examples for a specific use case.
Model announcements, research papers, and product updates. Follow this to know when new Claude models drop and what changed.
You want to stay up to date on new models and features.