Agent permission modes (YOLO mode)
Settings that decide what a coding agent may do without asking, from approving every command to skipping all checks, known as YOLO mode.
Coding agents ask before editing files or running commands, and that gets tiring fast. Permission modes trade safety for speed. Claude Code’s permission modes range from asking for everything, through auto-accepting edits and a read-only plan mode, to bypassing permissions, the mode people call YOLO; allow and deny rules for specific commands and paths sit on top.
Skipping prompts is reasonable only where the damage is contained: inside an agent sandbox or container with no production credentials and limited network. On a laptop with SSH keys and cloud access it turns a prompt injection into excessive agency. Teams usually set a shared default in repository settings and block the bypass mode on company machines.
Sources
- Claude Code Docs: Choose a permission mode code.claude.com
Related terms
- Agent sandbox: An isolated environment where an AI agent runs code and commands without access to your network, secrets or real files.
- Plan mode: A coding agent mode in which it can read code and run safe commands but not edit files, and ends with a plan for you to approve.
- Excessive agency: An AI agent with more tools, permissions or autonomy than its task needs, so a mistake or an injected instruction can do real damage.
- Human in the loop: A person approves, edits or rejects the agent’s action before it takes effect.
- Data classification gate for agents: A check that keeps secrets, personal data and other restricted data out of what a coding agent can read, send to a model or write to logs.
