Cheatsheet

I Want To…

I want to…	Do this
Start Claude Code	`claude`
Ask about a specific file	`claude "explain src/app.py"`
Fix a bug	Describe the error + paste the traceback
Create a new project	Describe what you want in plain English
Ask a one-off question without starting a session	`claude -p "what does this regex do: ^[a-z]+$"`
Resume where I left off	`claude --continue`
Commit my changes	`/commit`
Review code before committing	`/review`
Compress the context to save tokens	`/compact`
See how much I’ve spent this session	`/cost`
Clear the conversation and start fresh	`/clear`
Get help with available commands	`/help`
Run Claude without any interactive prompts	`claude --print -p "your task"`
Use a specific model	`claude --model claude-opus-4-5`
Limit how many steps Claude can take	`claude --max-turns 10`

Claude Code — CLI Flags

claude                          # Start interactive session
claude -p "prompt"              # One-shot, non-interactive (alias: --print)
claude --continue               # Resume last conversation
claude --model <model>          # Specify model (haiku/sonnet/opus or full name)
claude --max-turns <n>          # Limit agentic turns
claude --output-format json     # Machine-readable JSON output
claude --output-format text     # Plain text output (default)
claude --verbose                # Show full tool call detail
claude --allowedTools "Bash,Read,Write"  # Restrict which tools Claude can use
claude --no-tools               # Disable all tools (text only)
claude --add-dir /path/to/dir   # Add a directory to Claude's context

Keyboard Shortcuts (In-Session)

Shortcut	Action
`Ctrl+C`	Cancel current generation
`Ctrl+D`	Exit session
`Up arrow`	Navigate previous messages
`Shift+Enter`	New line in your message (without sending)
`Tab`	Autocomplete file paths
`Escape`	Cancel current input

Slash Commands (In-Session)

Command	What It Does
`/help`	Show available commands and current settings
`/status`	Session status, model in use, token count
`/cost`	Token usage and cost for this session
`/compact`	Summarize conversation to free up context window
`/clear`	Clear conversation history
`/memory`	View and edit CLAUDE.md memory files
`/commit`	Stage and commit changes with AI-written message
`/review`	Review current changes before committing
`/doctor`	Check Claude Code setup for common issues
`/model <name>`	Switch model mid-session

Prompt Patterns

One-liners for common tasks — copy, fill in the blanks, send.

Summarize the key points from [document/article] in bullet points.

Review this code for bugs, security issues, and readability. List issues by priority.

You are [role]. [Context]. Task: [what you need]. Format: [how you want it].

Explain [concept] like I understand [adjacent thing] but am new to this area.

I have [problem]. Here's what I've already tried: [attempts]. What should I try next?

Look at [file/folder] and tell me how it works before we make any changes.

Before writing any code, tell me your implementation plan. I'll approve before you start.

This is the output I got: [output]. This is what I expected: [expectation]. What's wrong?

Model Comparison

Model	Speed	Cost	Token Limit	When to Use
Claude Haiku 4.5	Fastest	Lowest (~$0.80/$4 per M)	200K	Classification, routing, high-volume simple tasks
Claude Sonnet 4.6	Fast	Mid (~$3/$15 per M)	200K	Most tasks — daily driver
Claude Opus 4.6	Slower	Highest (~$15/$75 per M)	200K	Complex reasoning, analysis, when quality is worth the cost

Input/output cost per million tokens. Check anthropic.com/pricing for current rates.

Rule of thumb: Default to Sonnet. Switch to Haiku when running many calls that don’t need high quality. Switch to Opus when you have a hard problem and you can tell Sonnet is struggling.

Token and Pricing Quick Reference

Concept	What It Means
Token	~4 characters of text (~0.75 words)
1,000 tokens	~750 words, or ~1.5 pages
Context window	How much text Claude can “see” at once (200K tokens = ~150K words)
Input tokens	Everything you send (system prompt + conversation history + your message)
Output tokens	Claude’s response — usually billed at a higher rate
Prompt caching	Reusing a long system prompt across calls — up to 90% savings on cached input

Rough cost estimates at Sonnet pricing:

Short Q&A (500 in / 300 out): ~$0.006
Summarize a long article (2K in / 500 out): ~$0.013
Analyze a 50-page document (40K in / 2K out): ~$0.15
Full codebase review (100K in / 5K out): ~$0.38

MCP Quick Reference

MCP (Model Context Protocol) servers give Claude access to external tools and data sources.

# List installed MCP servers
claude mcp list

# Add an MCP server
claude mcp add <name> <command>

# Remove an MCP server
claude mcp remove <name>

# Check MCP server status
claude mcp get <name>

Common MCP servers and what they do:

Server	What It Adds
`filesystem`	Read/write files anywhere on your system
`github`	Read PRs, issues, repos; post comments
`notion`	Read/write Notion pages and databases
`google-calendar`	Read/create calendar events
`gmail`	Read emails, create drafts
`firecrawl`	Scrape and read web pages as clean text
`sqlite`	Query a local SQLite database
`puppeteer`	Browser automation

API Quick Reference (Python)

import anthropic

client = anthropic.Anthropic()  # Uses ANTHROPIC_API_KEY env var

# Basic call
response = client.messages.create(
    model="claude-sonnet-4-5",
    max_tokens=1024,
    system="You are a helpful assistant.",        # optional
    messages=[{"role": "user", "content": "Hello"}]
)
print(response.content[0].text)

# Streaming
with client.messages.stream(
    model="claude-sonnet-4-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Write a story..."}]
) as stream:
    for text in stream.text_stream:
        print(text, end="", flush=True)

# Multi-turn conversation
messages = []
messages.append({"role": "user", "content": "What is machine learning?"})
response = client.messages.create(model="claude-sonnet-4-5", max_tokens=512, messages=messages)
messages.append({"role": "assistant", "content": response.content[0].text})
messages.append({"role": "user", "content": "Give me a concrete example."})
# Continue...

API Quick Reference (JavaScript)

import Anthropic from "@anthropic-ai/sdk";
const client = new Anthropic(); // Uses ANTHROPIC_API_KEY env var

const response = await client.messages.create({
  model: "claude-sonnet-4-5",
  max_tokens: 1024,
  system: "You are a helpful assistant.",
  messages: [{ role: "user", content: "Hello" }],
});
console.log(response.content[0].text);