Cline

Open-source AI coding agent for VS Code with Plan/Act modes

Free Dev Tools

Overview

Cline is an open-source AI coding agent that runs inside VS Code. It can edit files across your project, run terminal commands, plan before acting, and extend its capabilities through MCP servers. It supports every major AI model and is trusted by over 8 million developers.

⚡ Dev Tools 💵 Free - open source, bring your own API keys 📅 Listed 10 Jun 2026

Our take

Cline is an open-source coding agent for VS Code that runs on your own API keys, with explicit plan and act modes. Open source plus bring-your-own-key is a genuinely different proposition: you can read what it does, you pay per token rather than per seat, and your code goes to the provider you chose.

Worth it if

Developers who want an agentic assistant with visible behaviour and no subscription.

Look elsewhere if

Teams wanting managed support and billing rather than API keys and community support.

✨ Features

Plan/Act Modes

Plan mode analyzes tasks first, Act mode executes changes across your codebase.

Terminal Integration

Runs bash commands, installs packages, and manages your development environment.

Any Model Support

Use Claude, GPT, Gemini, local models, or any OpenAI-compatible API.

MCP Extensions

Extend with Model Context Protocol plugins for databases, APIs, and more.

⚖️ Pros & Cons

Pros

✓ Completely free and open-source
✓ Works with any AI model or API provider
✓ Deep VS Code integration with full project context

Cons

✗ Requires your own API keys (costs depend on model provider)
✗ Can be resource-intensive on large codebases

💰 Pricing

Free

Free

  • Fully open-source (Apache 2.0)
  • All features included
  • Use any AI model provider
  • MCP server support
  • Bring your own API keys

❓ Frequently Asked Questions

What do plan and act modes actually change?

Plan mode works out an approach before touching files; act mode executes it. The separation matters because agentic coding tools are most dangerous when they start editing before they understand the task. Reviewing the plan first is the cheapest way to avoid a large, confidently wrong change.

Is per-token cheaper than a coding assistant subscription?

For moderate use, usually. Agentic coding consumes more tokens than chat because it reads files repeatedly, so heavy use can exceed a subscription. Watch spend for the first fortnight — agentic token consumption is consistently higher than people expect before they measure it.