Plan mode
Setting that directs a coding agent to explore a codebase and propose changes before implementing them. In Claude Code, source edits normally remain blocked until the developer approves the proposed work.
In plain language
Review the proposed change before the agent starts implementing it.
In a sentence
- The developer selected plan mode to examine the proposed approach before implementation.
- In plan mode, the agent explored the codebase and drafted a plan for review.
Usage note
Planning does not imply a strict read-only environment. The agent may run shell commands while exploring, and permission exceptions depend on the tool.
Permission exception
Documentation explains how a product or system works. Claude Code documents an exception to plan-mode edit blocks in interactive terminal sessions where bypass permissions are available.
Other products
Documentation explains how a product or system works. Cursor's documentation says Plan Mode creates detailed implementation plans before writing any code.
See also
- Spec-driven development
Way of building software with AI in which a written specification comes before the code and serves as the source of truth for the human and the AI.
- Agent hook
User-defined action that a coding agent executes automatically at a specified point in its work, such as before or after a tool call.
- YOLO mode
Mode of running an AI agent with every action approved automatically, so its tool calls execute without asking the user for permission first.
- Background agent
Coding agent that works on a task in its own remote environment, such as an isolated virtual machine in the cloud, instead of on the developer's local machine, so the developer can delegate the task rather than pair with the agent in real time.
Cite this term.
CC BY 4.0, attribution required.
Orbyt Labs, Plan mode (AI Glossary), Nov 14, 2026. https://
Orbyt Labs Glossaries (Nov 14, 2026), https://
Last updated .