Back
WWindsurf logo

Windsurf

An AI code editor centered on the Cascade agent for multi-file understanding and edits.

AI codingCoding agentsFree plan
Official website

Plans and pricing

Official pricing verified
Access
Free plan
Paid plans from
$15/mo
Billing
Monthly subscription

Free

$0

Pro

$15/mo

Teams

$30/user/mo

Enterprise

Contact sales

Official pricing

DevPrice organizes public information and does not sell Windsurf subscriptions. Prices and availability are determined by Windsurf.

What is Windsurf?

Windsurf is an AI code editor built around Cascade as its primary interaction surface. Chat mode supports code explanation and planning, while Code mode can use tools, edit multiple files, and carry out development work. The goal is to keep codebase understanding, implementation, and result review inside the same editing environment.

Cascade can use project context from the editor, run terminal commands, and respond to code problems or linter output as it continues working. For features and refactors that cross file boundaries, it can locate relevant code and apply a sequence of changes. Checkpoints provide recovery points for larger edits, but users should still inspect the diff and run the repository's tests.

Windsurf uses Memories, Rules, and AGENTS.md files to preserve ongoing constraints, while Workflows and Skills package repeatable procedures. It can also connect to external tools through MCP. These capabilities help individuals and teams with established engineering practices, but higher autonomy makes explicit permissions, project rules, and validation commands more important.

Key features of Windsurf

  • Separate Chat and Code modes

    Chat is designed for questions and implementation discussion, while Code can invoke tools and modify the project. Users can switch between explanation and execution based on task risk.

  • Cross-file understanding and coordinated edits

    Cascade can inspect repository context to find entry points, dependencies, and call relationships before applying coordinated changes across several files.

  • Terminal and linter integration

    The agent can run terminal commands and respond to code issues detected by the editor, connecting dependency setup, scripts, and routine corrections in one workflow.

  • Checkpoints and change recovery

    Checkpoints can preserve a state before significant edits and restore it when the result goes off course, reducing the risk of manually untangling multi-step agent changes.

  • Memories, Rules, and project instructions

    Automatic memories preserve useful project context, while Rules and AGENTS.md let users define coding style, repository constraints, and execution requirements explicitly.

  • Reusable workflows and external tools

    Workflows and Skills package common procedures, while MCP connects external services and tools so established development routines can be repeated by Cascade.