How to Download and Install Devin AI CLI and Desktop

On June 2, 2026, Cognition announced the rebranding of the Windsurf IDE into Devin Desktop, representing a shift where 100% of local agent workflows are unified under the Agent Command Center rather than standalone local applications [Cognition 2026 Rebrand Announcement]. While Stack Overflow's 2026 Developer Survey reporting that over 84% of software developers now integrate AI agents into their daily programming workflows [Stack Overflow 2026 Survey], the key operational bottleneck has shifted from basic code generation to the complex synchronization of local files and cloud sandboxes. This guide provides the exact steps to download, install, and configure Devin AI for local work while connecting to your remote environments.

Developers require environments that combine the convenience of local terminals with the scaling advantages of remote execution. The Devin CLI and Devin Desktop allow developers to run coding agents locally while handing off complex workloads to the Devin Cloud environment. Devin CLI runs locally on macOS, Linux, and Windows WSL, whereas Devin Desktop serves as the Agent Command Center, formerly known as Windsurf. By decoupling the user interface from the container where execution actually happens, developers can maintain their local editor configurations, extensions, and personal settings, while allowing the agent to run long-running tasks inside a remote, isolated sandbox. This hybrid architecture preserves battery life, limits local processor load, and allows for asynchronous runs that continue even if the developer's local machine goes offline.

Installing Devin Desktop provides developers with a full development environment equipped with agent command tools. For developers transitioning from Windsurf, the change to Devin Desktop occurs as an automated update. The update retains all active user extensions, keybindings, settings, and workspace histories, migrating them to the new environment.

If you are performing a fresh installation, download the package from the official download URL at https://devin.ai/desktop. Follow these instructions to install the desktop application:

1. Open your browser and navigate to the official desktop page.

2. Select the installer compiled for your operating system. The platform supports macOS (Universal binary), Windows (EXE installer), and Linux (AppImage or DEB package).

3. Open the downloaded file. On macOS, mount the disk image and drag the application to your Applications directory. On Windows, run the executable and follow the prompts to place the files in your program directory.

4. Launch the application. You will be redirected to your browser to complete a sign-in flow. Log in with your Cognition credentials to link the desktop app to your account.

Once authenticated, the IDE launches the Agent Command Center. This interface features a Kanban view where you can monitor active agents, review pull requests, and organize multiple sessions. Devin Desktop supports the Agent Client Protocol (ACP), which allows you to run other models and agents alongside Devin, providing a unified command interface for all your automation tools.

The Devin AI CLI provides a terminal interface for developers who prefer to trigger runs, manage files, and automate tasks from the command line. The installation relies on direct shell scripts hosted on the official cli domain.

To install the CLI on macOS, Linux, or Windows WSL, run the following command in your terminal:

curl -fsSL https://cli.devin.ai/install.sh | bash

If you are running a native Windows environment, open PowerShell with administrator privileges and execute:

irm https://static.devin.ai/cli/setup.ps1 | iex

The installation script downloads the binary, moves it to your system path, and sets up shell completions for bash, zsh, and fish. After the script completes, close and reopen your terminal to reload your shell environment. Verify the command availability by running:

devin --version

To authenticate your terminal session with the Cognition cloud, execute the login command:

devin login

This action opens your default web browser and prompts you to authorize the CLI. Once approved, the CLI writes a configuration file to your home directory under the .devin folder. This token authorizes the local CLI to launch remote virtual machines, monitor jobs, and download results from cli.devin.ai and docs.devin.ai. You can customize default settings, such as remote memory allocation and max run time, by editing the .devin/config.json file.

Most articles focus purely on the cloud-based Devin web interface and miss explaining how to configure the local CLI and manage persistent files across local and cloud sandboxes. This oversight leaves developers struggling to sync datasets, assets, and build caches. A remote Devin session runs in an isolated container that is destroyed when the task finishes. While Git is the default method for saving code changes, it is not designed to handle non-code assets. Developers cannot commit database dumps, test logs, media files, or large data exports to Git without bloating the repository.

To resolve this sandbox gap, teams use different file storage patterns:

1. Local machine directories: Developers can sync folders using local file-watching daemons. However, this approach consumes local bandwidth and fails if the developer's computer goes offline or loses connection during a run.

2. Cloud object storage: Writing upload scripts to transfer logs and assets to Amazon S3 is a common alternative. However, this requires writing upload code, managing API keys inside the agent environment, and lacks version history and search tools.

3. Fastio workspaces: Pairing Devin with Fastio workspaces (details at /product/workspaces/) offers a more integrated environment. Fastio provides shared, organization-owned workspaces where humans and agents collaborate on the same files. Fastio workspaces feature per-file version history, ensuring that if an agent writes a file or updates a script, the complete change history is tracked. If multiple agents or human developers write to a file at the same time, earlier versions can be restored. Fastio manages these file structures and access histories without claiming compliance certifications. Fastio supports granular permissions, ensuring that sensitive project folders are only accessible by authorized service users or human team members. Fastio provides these workspace management capabilities directly without claiming compliance certifications.

Store Devin AI session output persistently

Configure a shared team workspace for your developer agents with persistent directories, file version history, and automatic semantic indexing. Starts with a 14-day free trial.

Start 14-Day Trial

Pairing Devin with Fastio turns basic storage into an intelligent workspace. The integration is established using the Model Context Protocol (MCP). Fastio exposes a consolidated MCP toolset through Streamable HTTP at /mcp and legacy SSE at /sse. By connecting the agent to the Fastio MCP server, developers enable Devin to read and write files directly within shared team folders. The tool configurations and setup steps are available in the Fastio MCP server guide, and agent onboarding documentation can be accessed at https://fast.io/llms.txt.

Once the integration is active, several features improve the development workflow:

Intelligence Mode: Enabling this mode on your workspace directs Fastio to index all files for semantic search. Instead of searching by exact filename, Devin can perform semantic queries to locate code, specifications, or media assets by meaning, returning answers with citations to the exact source files.

Metadata Views: Turn documents into a live, queryable database. By visiting the Metadata Views page, you can define columns in plain English (such as client name, invoice amount, or build date). Fastio automatically extracts this structured data from PDFs, images, and text files. Devin can query these spreadsheet-style grids via MCP, allowing the agent to read structured document fields.

Handoff Workflows: Devin can build code, generate reports, and place them in a workspace. When the task is complete, the agent can transfer workspace ownership to a human manager. Fastio does not offer a permanent free plan or free agent tier. The platform requires a paid subscription starting with a 14-day free trial that requires a credit card. Plans include Starter at $29/mo, Business at $99/mo, and Growth at $299/mo. Humans can register their agents, configure shared folders, and evaluate the tools during the trial.

We should note that public file sharing is supported using Receive and Exchange shares to collect outbound and inbound files, making it simple to collect results from external systems. Fastio workspaces default to intelligence enabled when created by agents, which helps automate the RAG setup for any new workspace folder.

Setting up developer tools locally often introduces path and credential errors. Below are the steps to resolve the three most common setup bottlenecks:

Windows Execution Policies

If PowerShell blocks the Devin setup script, it is due to restrictive execution policies. To bypass this, run PowerShell as administrator and set the policy temporarily for the current session:

Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope Process

After setting the policy, run the install script again.

WSL Path Resolution

When installing the CLI inside WSL, the command path may not reload automatically. If the terminal returns a command not found error after installation, force a shell refresh:

source ~/.bashrc

This reload ensures that the directory where the binary was placed is added to the system path.

Browser Redirect Failures

In headless environments or restricted terminals, the authentication browser may fail to redirect back to the CLI. If this occurs, copy the authorization URL from the terminal, open it in your browser, log in, and copy the manual auth token. Paste the token back into the terminal prompt to complete the setup.