Cline

Open-source AI coding agent for your editor, terminal, and apps

Intermediate API

What is Cline?

Cline is an open source AI coding agent available as an editor extension, terminal CLI, desktop app, and embeddable SDK. It supports Plan and Act modes, multi-file edits with undo, any major model provider including local weights, and automation through MCP, chat apps, and CI pipelines.

Cline is an open source AI coding agent that runs across three surfaces: an editor extension, a terminal CLI, and a native desktop app. Its core promise is model neutrality. Instead of locking users to one provider, Cline connects to Claude, GPT, Gemini, local Ollama or LM Studio instances, or any OpenAI-compatible endpoint, so teams can bring their own API key or their own weights without changing how the agent behaves. The project reports adoption by more than 11 million developers. The workflow is built around Plan and Act modes. Plan mode is for aligning on a strategy before any files are touched, and Act mode carries out the work. Every step can require explicit approval, or auto-approve can be enabled for unattended runs. Edits span multiple files with linter-aware fixes, diffs, checkpoints, and one-click undo, which matters when an agent is making coordinated changes across a real repository. Cline also runs bash commands and reacts to live output, so long-running test suites, development servers, and deploy scripts can be handled inside the session rather than pasted between tools. Extensibility is where the product separates from simpler editor assistants. A .clinerules file ships with a repository to teach the agent coding standards, architecture, and deployment conventions, while Skills and MCP servers add database, API, and infrastructure access. The SDK exposes the same runtime for embedding in other applications, and coordinator agents can delegate to specialist agents with their own tools and context, including on a cron schedule. Cline can be driven from Slack, Discord, Telegram, or Linear, and run headlessly inside GitHub Actions, GitLab, or other CI pipelines. The audience is broad: individual developers who want a capable agent inside VS Code, Windsurf, Cursor, or JetBrains; teams that standardize coding conventions through repository rules; and platform engineers who need an agent runtime they can embed or automate. Cline competes with closed, single-model coding assistants by trading some setup effort for transparency, provider choice, and self-hosting options.

How do you use Cline?

  1. 1Choose your surface: IDE, terminal, or desktop
    Decide where the agent should run. The editor extension suits interactive work, the CLI suits terminal-first workflows, and the desktop app works without an editor.
  2. 2Install the extension or CLI
    Install Cline from the VS Code, Windsurf, Cursor, or JetBrains marketplace, or run npm i -g cline for the terminal version.
  3. 3Connect a model provider
    Add an API key for Claude, GPT, Gemini, or another OpenAI-compatible endpoint, or point Cline at a local Ollama or LM Studio instance.
  4. 4Plan the task before acting
    Describe the change in Plan mode to align on strategy, then switch to Act mode and approve steps individually or enable auto-approve.
  5. 5Add repository rules and tools
    Commit a .clinerules file to teach the agent your standards, and register MCP servers for databases, APIs, or infrastructure access.
  6. 6Embed or automate with the SDK
    Install @cline/sdk to embed the runtime in your own apps, or run the agent headlessly inside CI and chat integrations.

Pros and cons

Pros

  • Works with any major model provider, local Ollama or LM Studio, and OpenAI-compatible endpoints, so no vendor lock-inAI
  • Plan and Act modes let users review strategy before execution, with per-step approval or auto-approveAI
  • Multi-file edits include diffs, checkpoints, and one-click undo, which limits damage from bad agent stepsAI
  • Runs bash commands and reacts to live output for tests, dev servers, and deploysAI
  • SDK and MCP support allow embedding the runtime and connecting external tools and data sourcesAI

Cons

  • Bring-your-own-key model means users pay providers separately and must manage credentials and rate limitsAI
  • Powerful automation features like coordinator agents and headless CI runs require significant setup knowledgeAI
  • The number of configuration surfaces (IDE, CLI, desktop, SDK) can be confusing for newcomers choosing where to startAI
  • No bundled model means output quality and cost vary widely depending on which provider or local model is connectedAI

How much does Cline cost?

Support

Full documentation is available at docs.cline.bot, covering setup, rules, MCP, and the SDK. The project is open source, so code and releases are public.

Learn more

Integrations

Connects to Claude, GPT, Gemini, local Ollama and LM Studio, and any OpenAI-compatible endpoint. Chat integrations cover Slack, Discord, Telegram, and Linear, and headless runs work inside GitHub Actions, GitLab, and other pipelines. MCP servers add databases, APIs, and infrastructure.

Learn more

Features

Plan and Act modes with per-step approval or auto-approve; coordinated multi-file edits with linter-aware fixes, diffs, checkpoints, and one-click undo; bash command execution with live output handling; .clinerules and Skills for repository conventions; MCP servers and SDK lifecycle hooks; Slack, Discord, Telegram, and Linear chat access; headless CI runs; coordinator agents with cron schedules.

Learn more

Frequently asked questions about Cline

  • Does Cline have an API?
    Yes, Cline offers an API.
  • How do you use Cline?
    The walkthrough on this page covers 6 steps: 1. Choose your surface: IDE, terminal, or desktop 2. Install the extension or CLI 3. Connect a model provider 4. Plan the task before acting 5. Add repository rules and tools 6. Embed or automate with the SDK.
  • What platforms does Cline support?
    Cline is available on Web App, MacOS, Windows, Linux and Browser extension.
  • What does Cline integrate with?
    Connects to Claude, GPT, Gemini, local Ollama and LM Studio, and any OpenAI-compatible endpoint. Chat integrations cover Slack, Discord, Telegram, and Linear, and headless runs work inside GitHub Actions, GitLab, and other pipelines. MCP servers add databases, APIs, and infrastructure.
  • What are the limitations of Cline?
    Bring-your-own-key model means users pay providers separately and must manage credentials and rate limits. Powerful automation features like coordinator agents and headless CI runs require significant setup knowledge. The number of configuration surfaces (IDE, CLI, desktop, SDK) can be confusing for newcomers choosing where to start.

Status

StatusActive
Views0
Outbound clicks0
Added10/21/2026

Platforms

Web AppMacOSWindowsLinuxBrowser extension

Pricing

Free