Default & Plan Modes: Safety First

⏱ Est. reading time: 3 min Updated on 5/8/2026

These modes are designed for exploration and maximum human oversight.

Default Mode

Behavior: Only "built-in read-only" tools run without asking. Everything else prompts for approval.

  • Read-only Whitelist: ls, cat, grep, find, diff, stat, du, cd, and read-only git commands.
  • Access: This is the initial state when you start Claude Code.
  • Indicator: No special indicator in the status bar.

Plan Mode

Behavior: Claude can read files and explore, but cannot modify source code. It focuses on creating a plan.

Acceptance Options

Once a plan is ready, Claude offers several options:

  1. Approve and start in auto mode
  2. Approve and accept edits
  3. Approve and review each edit manually
  4. Keep planning with feedback

Tip: Use /plan <task> to enter plan mode and start immediately (e.g., /plan fix auth bug).

Related Tools & Resources

Skill Marketplaces

A

Awesome Cyber Skills

A curated list of hacking environments for AI agent developers to train and practice cybersecurity skills, enhancing agent capabilities in defense and offense.

Recommended Plugins

🛡️

Security Guidance

Security reminder hook that monitors 9 security patterns including command injection, XSS, eval usage, dangerous HTML, pickle deserialization, and os.system calls.

Related Products

📦

claude-howto

Claude How To by luongnv89 is a structured, visual, and example-driven guide designed to help users master Claude Code's features efficiently. It offers 10 tutorial modules, copy-paste configuration templates, Mermaid diagrams, and a guided learning path. The guide transforms users from basic Claude Code interactions to orchestrating advanced agents, hooks, skills, and MCP servers, addressing the gap in official documentation regarding practical guidance and a clear learning roadmap.