Coding with AI tools

Stage 7 · Lesson 2 of 5

Permission modes and rules in Claude Code

Agents: multi-file changes, permissions and safety

's decides what runs without asking. Manual mode, whose config value is default, asks before edits and commands. acceptEdits lets file edits and simple file commands through.

plan reads and plans only. auto has a separate model review actions and block risky ones; recent versions start in it by default. dontAsk denies anything not pre-approved, for scripts.

bypassPermissions skips checks and belongs only in isolated containers or VMs. Press Shift+Tab to cycle modes. On top, in .claude/settings.json, such as Bash(npm run test *) or Read(./.env), pre-approve or block specific actions, and deny rules apply in every mode. /permissions manages them.

Common trapallowing a broad pattern, such as any command, which also allows installing any package.

A question to check yourself, an AI practice session and 2 cheat sheet entries come when you add the roadmap to your account. Add it

Resources for this stage

Start with· follow one of these

  • GuideFree
    Building Effective Agents

    Anthropic · Advanced

    Anthropic's clear, free essay on how agents work and when simpler workflows are better

  • CourseFree
    Claude Code in Action

    Anthropic · Intermediate

    Free course lessons on permission modes, hooks and verifying unsupervised runs

Keep handy

Also worth your time

Mark this stage as not done?