Cline

Open-source VS Code agent that plans, edits and runs commands on your key

AI Coding & DevelopmentFree planOverseasβ˜…β˜…β˜…β˜…β˜… 5.0

What is Cline?

Cline is an open-source coding agent that runs as an extension inside your editor. It splits work into a planning step and an acting step, so you can see and approve the approach before it touches files. It edits across the project, runs terminal commands, uses a browser to check its own output, and takes checkpoints so any change can be rolled back. Model access is bring-your-own-key: Anthropic, OpenAI, Google, OpenRouter or a local runtime all work. The differentiator is that the agent itself is free and auditable, and you decide which model and budget it runs against.

Last updated: 2026-09-20. This site only provides an index; for exact features, pricing, and licensing, see the official website.

Key features

  • Plan and act modes that separate proposals from executed changes
  • Edits multiple files and runs terminal commands in your workspace
  • Checkpoint and rollback for every change it makes
  • Browser automation so it can verify its own web output
  • MCP marketplace for adding tools and data sources
  • Bring your own key for Anthropic, OpenAI, Google, OpenRouter or local models

Pros & cons

Strengths

  • Free and open source with fully auditable behaviour
  • You control model choice and therefore cost per task
  • Checkpoints make risky edits easy to undo

Watch out for

  • Requires an API key, so there is a running model cost
  • Long autonomous runs can consume tokens quickly
  • Agent quality tracks whatever model you connect to it

Best for & use cases

agentic editing, byok workflows, refactoring, automated bug fixing

If you're comparing similar products, check the alternatives below, or browse all tools in the AI Coding & Development category.

FAQ

Is Cline free?

The extension is free and open source. Your real cost is the model tokens it consumes through whichever provider you connect, unless you run a local model, in which case there is no per-request charge.

How does plan and act mode help?

Planning mode lets the agent investigate and propose an approach without changing files. You review that plan, then switch to act mode so it executes. It keeps surprises low on unfamiliar codebases.

Can it use a local model?

Yes, it can point at local runtimes that expose an OpenAI-compatible API such as Ollama or LM Studio. Results depend on how capable a model you can comfortably run locally.