Claude Code
- Category
- Terminal AI Coding Agent
- Platform
- macOS, Windows, Linux, VS Code extension
- Pricing
- Pro ($20/mo) / Max 5x ($100/mo) / Max 20x ($200/mo)
- Models
- Claude Opus 4.7+, Sonnet 4.6, Haiku
- Context File
- CLAUDE.md + AGENTS.md
- Key Features
- Subagents, session forking, MCP, hooks, sandbox, /compact
Claude Code is Anthropic's official terminal agent — built for multi-file refactors, long agentic loops, and repo-wide reasoning. It runs in your shell (or VS Code) with full filesystem and terminal access, and has become the default CLI for developers who want frontier reasoning without leaving the terminal.
Why Web Devs Reach for It
Cursor wins daily inline edits; Claude Code wins when the task spans dozens of files, needs autonomous command execution, or requires the strongest available reasoning model. Web projects benefit from its first-class MCP support (Playwright, Figma, GitHub), permission profiles, and CLAUDE.md for framework-specific rules.
2026 Differentiators
- Subagents — delegate research, tests, or refactors to specialized child agents
- Session forking — branch a conversation to explore alternatives without losing context
- Auto mode — let the agent pick the right model per task
- Hooks — run scripts before/after tool calls for linting, tests, or secret scanning
- MCP ecosystem — connect browser, database, and design tools natively
Honest Limits
No built-in tab completion — pair it with Cursor or Copilot for inline flow. Token costs add up on Max tiers during long sessions; use /compact and subagents deliberately. Requires an Anthropic subscription; no BYOM like OpenCode or Cline.