Frequently Asked Questions

Quick answers to common questions about Claude Code. For detailed guides, follow the links to relevant documentation sections.

Product Scope & Clarity

Q: Is this guide for Cursor IDE?

A: No. Cursor is a separate third-party IDE product. This guide covers Claude Code (official Anthropic tools: CLI + VS Code Extension). We have a separate guide for Cursor.

Q: What’s the difference between Claude Code and Cursor?

A: They are completely different products:

Feature Claude Code (This Guide) Cursor IDE (Separate Product)
What is it? Official Anthropic CLI + VS Code Extension Third-party AI-powered IDE
Maker Anthropic (official) Cursor (third-party)
Installation curl https://claude.ai/install.sh \| bash Download Cursor app
Works with Any editor (CLI), VS Code (extension) Cursor IDE only
Best for Automation, CI/CD, scriptable workflows Full IDE experience

Q: Can I use both Claude Code and Cursor?

A: Yes, but choose one for IDE integration:

  • Claude Code CLI: Works regardless of your editor (use with any IDE)
  • For IDE integration: Choose either Claude Code VS Code Extension OR Cursor IDE (not both in VS Code)

Q: Which should I use - Claude Code or Cursor?

A: Depends on your needs:

Use Claude Code if you want:

  • ✅ Official Anthropic tools
  • ✅ Terminal automation and scripting
  • ✅ CI/CD integration
  • ✅ Works with your existing editor
  • ✅ More control over prompts and models

Use Cursor if you want:

  • ✅ Fully integrated IDE experience
  • ✅ More powerful multi-file editing
  • ✅ Built-in AI pair programming
  • ✅ Don’t want to learn CLI commands

Many developers use both: Claude Code CLI for automation + Cursor for daily coding.

Q: This guide mentions “VS Code Extension” - what is that?

A: The Claude Code VS Code Extension is the official Anthropic extension for Visual Studio Code. It’s part of Claude Code (not Cursor). Install it from VS Code Marketplace by searching “Claude Code”.

Getting Started Questions

Q: Do I need both the CLI and VS Code Extension?

A: No, choose based on your workflow:

  • CLI only: Perfect for terminal users, automation, CI/CD
  • Extension only: Perfect for GUI-first developers
  • Both: Maximum flexibility (recommended)

Q: How much does Claude Code cost?

A: Claude Code itself is free. You pay for Claude API usage:

  • Typical individual developer: $5-10/month (optimized)
  • Heavy user: $20-50/month
  • Team of 10: $50-200/month (optimized)

See Cost Optimization Guide for detailed strategies.

Q: Can I use Claude Code without internet?

A: No, Claude Code requires API access to Anthropic’s servers. However, you can:

  • Cache context locally (faster responses)
  • Use Batch API for async processing
  • Self-host via AWS Bedrock or GCP Vertex (enterprise)

Q: What’s the difference between Claude API and Claude Code?

A:

  • Claude API: Raw API access, requires programming
  • Claude Code CLI: Terminal interface to Claude (easier than raw API)
  • Claude Code VS Code Extension: GUI interface in VS Code
  • Agent SDK: Build custom agents programmatically

All use Claude API underneath.

Technical Questions

Q: Which model should I use?

A: Quick guide:

  • Simple tasks: Haiku 4.5 (fast, cost-effective)
  • Daily coding: Sonnet 4.5 (balanced)
  • Complex architecture: Opus 4.5 (powerful)

See Mental Models & Architecture for detailed decision matrix.

Q: What’s extended thinking and when should I use it?

A: Extended thinking lets Claude “think through” problems before responding (1K-32K tokens of reasoning).

Use when:

  • ✅ Complex architecture decisions
  • ✅ Debugging intricate issues
  • ✅ Multi-step analysis

Don’t use when:

  • ❌ Simple questions
  • ❌ Code formatting
  • ❌ Quick reviews

Cost impact: 2-3x more expensive, but 10x better for complex tasks.

Q: How do I reduce costs?

A: Top 3 strategies:

  1. Create CLAUDE.md → 90% cost reduction via caching
  2. Use Haiku for simple tasks → 10x cheaper than Opus
  3. Enable Batch API → 50% discount for async work

See Cost Optimization Guide for complete strategies.

Q: What’s CLAUDE.md and do I need it?

A: CLAUDE.md is a project context file that:

  • Describes your project architecture
  • Lists conventions and patterns
  • Defines common tasks

Benefits:

  • Claude uses it automatically (no manual attachment needed)
  • Enables prompt caching (90% cost savings)
  • Consistent responses across sessions
  • Great for onboarding

Required? No, but strongly recommended for multi-session projects.

See CLAUDE.md Guide for templates and examples.

Troubleshooting Questions

Q: Why is Claude Code slow?

A: Common causes:

  1. Extended thinking enabled → Disable for simple tasks
  2. Large context → Narrow file selection
  3. No streaming → Enable streaming: --stream
  4. Using Opus → Switch to Haiku for speed

See Performance Optimization for detailed solutions.

Q: Why are my costs so high?

A: Check these:

  1. No prompt caching → Create CLAUDE.md
  2. Using wrong model → Use Haiku for simple tasks
  3. Re-reading files → Claude caches, don’t re-attach
  4. Extended thinking always on → Only use for complex tasks

Run: claude --usage-stats --month=current to analyze.

See Cost Optimization for detailed strategies.

Q: Authentication failed - what do I do?

A: Steps to fix:

# 1. Check if key is set
echo $ANTHROPIC_API_KEY

# 2. If empty, set it
export ANTHROPIC_API_KEY="your-key-here"

# 3. Verify it works
claude "test" --verbose

# 4. Make permanent (add to ~/.bashrc or ~/.zshrc)
echo 'export ANTHROPIC_API_KEY="your-key-here"' >> ~/.bashrc

Get key from: https://console.anthropic.com/

Q: VS Code Extension not showing up?

A: Checklist:

  • VS Code version 1.98.0+ required
  • Extension installed: Search “Claude Code” in Extensions
  • Reload window: Cmd/Ctrl+Shift+P → “Reload Window”
  • API key configured in extension settings
  • Check extension logs for errors

See Troubleshooting Guide for comprehensive solutions.

Workflow Questions

Q: Should I review Claude’s code before committing?

A: YES, ALWAYS. Claude is a tool, not a replacement for judgment:

  • ✅ Review all generated code
  • ✅ Test thoroughly
  • ✅ Check for security issues
  • ✅ Ensure it matches your patterns
  • ❌ Never auto-commit without review

See PR Review Guide for AI code review checklist.

Q: Can Claude Code work in CI/CD?

A: Yes! Perfect use cases:

  • Automated code reviews on PRs
  • Test generation
  • Documentation generation
  • Security scanning
  • Release notes generation

See DevOps Integration for setup guides.

Q: How do I share prompts with my team?

A: Best practices:

  1. Shared prompt library: Create docs/claude-prompts.md
  2. Team CLAUDE.md template: Version-controlled project templates
  3. Custom skills: Build team-specific skills
  4. Wiki documentation: Internal knowledge base

See Team Collaboration for team patterns.

Q: Can I use Claude Code for non-code tasks?

A: Yes! Great for:

  • Writing documentation
  • Analyzing logs
  • Data transformation
  • Content generation
  • Technical writing
  • Architecture decisions

Just pipe any text to Claude: cat log.txt | claude "summarize errors"

Enterprise & Security Questions

Q: Is Claude Code secure for enterprise use?

A: Security features:

  • ✅ SOC 2 Type II certified
  • ✅ Data encryption in transit and at rest
  • ✅ No training on your data (by default)
  • ✅ Self-hosted options (AWS Bedrock, GCP Vertex)
  • ✅ Audit logging available

See Security & Compliance for details.

Q: Can we self-host Claude Code?

A: Sort of. Options:

  1. CLI/Extension: No self-hosting (use API)
  2. API via AWS Bedrock: Your AWS account, your VPC
  3. API via GCP Vertex: Your GCP project
  4. On-premises: Not currently available

AWS/GCP hosting gives you data control without managing models.

Q: How do we manage API keys for a team?

A: Recommended approaches:

  1. Individual keys: Each developer has own key (tracked separately)
  2. Team key: Shared key in secure secret manager (AWS Secrets Manager, 1Password)
  3. Enterprise: Centralized billing, individual attribution

Never: Commit keys to git, share via Slack, put in code.

Comparison Questions

Q: Claude Code vs GitHub Copilot?

A: Different strengths:

Feature Claude Code GitHub Copilot
Autocomplete ❌ No ✅ Excellent
Multi-file refactoring ✅ Excellent ❌ Limited
Context awareness ✅ 200K-2M tokens ⚠️ Limited
Reasoning ✅ Extended thinking ❌ No
CLI automation ✅ Native ❌ No
Cost Pay per use Flat $10-20/mo

Many developers use both: Copilot for autocomplete, Claude Code for complex tasks.

Q: Can Claude Code replace code review?

A: No, it augments it:

  • ✅ Use Claude to catch obvious bugs before human review
  • ✅ Use for security/performance first-pass
  • ✅ Frees humans to focus on architecture/design
  • ❌ Don’t skip human review
  • ❌ Don’t blindly trust AI suggestions

Best practice: Claude pre-review → Human review → Merge

Still Have Questions?

Ask in the community:

Or search this guide:

← Back: PR Review Guide Up: Part 7 Index
This site uses Just the Docs, a documentation theme for Jekyll.