How to Set Up Cline in Visual Studio Code

As of July 2026, the Cline extension for VS Code reports over 4,500,000 installs on the Visual Studio Marketplace [VSMarketplace 2026 stats]. This high volume reflects a massive shift in how developers write code. While early AI assistants focused on simple code completion, modern developer teams are adopting autonomous agents that can edit files, run tests, and execute shell commands inside the local workspace. Cline is provider-agnostic and supports local/remote LLMs, running directly inside your editor to execute complex engineering tasks under your direct supervision.

However, running an autonomous agent locally creates challenges for data management and collaboration. By default, all file edits, terminal transcripts, and code changes are stored on your local hard drive. If you switch from your office computer to a laptop at home, or if you collaborate with a team, your agent's files and operational history remain locked to that single machine. Traditional alternatives like Git branches, S3 buckets, or consumer cloud folders present problems. Git branches require manual commits and merges, S3 buckets require complex credentials and lack a user interface, and consumer folders do not support semantic search or real-time document extraction.

To solve this, developers are connecting local agents to shared cloud environments. A persistent cloud workspace acts as a central repository where humans and AI agents share files, database schemas, and workflows. Instead of keeping files on local disks, you can direct your agent to output its code, documentation, and test reports to a persistent workspace. This keeps your agent's files accessible across devices and allows human teammates to review results in real time. Developers can start syncing their workspaces with a 14-day free trial on the Fastio pricing page.

Setting up the extension is the first step toward local agentic coding. To complete the installation, follow these numbered steps:

1. Open Visual Studio Code on your machine.
2. Access the Extensions view by clicking the Extensions icon on the Activity Bar or pressing Command+Shift+X on macOS or Control+Shift+X on Windows.
3. Search for "Cline" in the search box at the top of the panel.
4. Click the install button on the official extension (extension ID: saoudrizwan.claude-dev).
5. Wait for the installation to complete, then confirm that the Cline robot icon appears on the Activity Bar.

Once the extension is installed, you can access its chat panel by clicking the robot icon. If you prefer running commands from your system terminal, you can install the CLI version of Cline. Running the command-line CLI version of Cline requires Nodejs 20+ to execute local file and agent commands [Cline Docs 2026]. You can verify your system's Nodejs version by running node -v in your command line before installing the CLI globally.

To configure Cline, you must choose an AI model provider and set up the corresponding credentials. Access the settings panel by clicking the gear icon in the top-right corner of the Cline sidebar window. Cline supports Anthropic Claude, OpenAI, OpenRouter, Google Gemini, AWS Bedrock, and local models.

When you select a commercial provider, you must input your API key. Understanding where Cline stores this data is essential for security:

• Encrypted secrets: Cline stores sensitive credentials, including API keys, in Visual Studio Code's secure secret storage. This system keeps your keys encrypted and prevents them from appearing in plain text.
• Global configuration files: General settings, such as your selected provider and custom system rules, are saved in the ~/.cline/data/settings/ directory on your machine.
• VS Code extension storage: The extension also writes operational metadata to the local global storage folder under globalStorage/saoudrizwan.claude-dev/settings/.

You can configure custom system instructions inside the settings panel to define how the agent behaves. These rules are stored globally, but you can override them on a per-project basis. Creating a .cline/ directory at the root of your project repository lets you store project-specific guidelines that the agent reads automatically. This configuration structure ensures your API keys stay secure while allowing you to customize the agent's behavior for different codebases.

Many development teams avoid commercial APIs due to data privacy policies or usage costs. Configuring Ollama allows you to run language models locally on your hardware, ensuring that your code files never leave your machine. Most articles omit local model configuration or the configuration of global state files, leaving developers to troubleshoot connection issues on their own.

Follow these steps to run Cline with Ollama:

1. Download and install the application from the official Ollama website.
2. Open your system terminal and pull a programming model, such as ollama pull qwen-coder.
3. Open the Cline settings panel and select Ollama from the API Provider dropdown menu.
4. Verify that the Base URL is set to the default address: http://localhost:11434.
5. Select your pulled model from the model list.

When running local models, you must increase the context window size to prevent execution issues. By default, local models use a small context window, which can cause Cline to run into infinite loops or lose track of files during complex tasks. You can adjust the context window size (num_ctx) to 32k or higher by creating a custom Modelfile for Ollama [Ollama Context 2026]. You can inspect the active model configuration and global settings inside the ~/.cline/data/settings/providers.json file to verify that the local connection parameters are configured correctly.

Sync Cline workspace files across machines

A shared workspace with an MCP-ready endpoint for your Cline agent's reads and writes, featuring automatic version history and semantic search. Starts with a 14-day free trial.

Start 14-Day Trial

For developers working across multiple machines or in teams, local storage creates barriers. You can connect Cline to a persistent cloud workspace using the Model Context Protocol (MCP). Fastio provides persistent storage and shared workspaces built for agentic workflows, offering a secure middle layer between your local IDE and cloud resources.

To connect Cline, you must edit the cline_mcp_settings.json file. You can open this file by clicking the MCP Servers icon in the Cline panel and selecting the configure option. Add Fastio under the mcpServers configuration block using the native Streamable HTTP transport:

{
  "mcpServers": {
    "fastio-storage": {
      "type": "streamableHttp",
      "url": "https://mcp.fast.io/mcp/key",
      "headers": {
        "Authorization": "Bearer your_api_token_here"
      },
      "disabled": false,
      "autoApprove": []
    }
  }
}

Using this connection gives your agent access to advanced cloud storage features:

• Intelligence Mode: Fastio automatically indexes all files written by the agent for semantic search and citation-backed RAG chat. You do not need to configure a separate vector database.
• Hybrid Search: The agent can run search queries that combine exact full-text matching with semantic meaning retrieval to locate code and documentation across folders.
• Version History: Every file written to the workspace maintains a detailed history. If Cline writes a bug or deletes a file during an autonomous run, you can restore previous versions immediately.

By offloading file processing, indexing, and search queries to the cloud, you reduce the CPU load on your local development machine, which is particularly beneficial when running local models like Ollama. Developers can learn more about configuring these parameters by checking the Fastio agent storage options or reading the onboarding instructions.

Modern software development relies on teamwork. When Cline completes a task, you must share code, documentation, and reports with human colleagues or clients. Fastio provides the coordination layer where agent output becomes team output, placing humans and AI agents on the same active canvas.

Humans can view files and collaborate in real time using the web UI, while agents interact via the Fastio API or MCP. Developers and agents can co-edit requirements, design docs, and tasks using Collaborative Notes, which features Google-Docs-style co-editing with live multiplayer cursors. If you need to share files with external QA teams, you can build branded shares (Send, Receive, or Exchange) that support custom logos, expiration parameters, and password access.

To turn unstructured agent files into a queryable database, teams can use Metadata Views. This feature allows you to define columns in plain English (such as "File Type," "Security Level," or "Last Author"). The AI designs a typed schema using Text, Integer, Decimal, Boolean, URL, JSON, or Date & Time field types, matches files in the workspace, and populates a filterable spreadsheet. Both humans and agents can query these views to track progress, as detailed on the Metadata Views product page.

When your development project is complete, you can hand off the workspace to your client or manager. The agent-developer flow allows you to sign up free, build the workspace, and transfer the organization to a human who joins and starts the 14-day free trial. Activated trials require a credit card. Fastio plans include Starter at $29/mo, Business at $99/mo, and Growth at $299/mo, giving teams a persistent workspace to manage their projects.