Code Execution

Code Execution

CodingFleet provides an integrated Code Execution feature that allows AI-generated code to run securely inside isolated sandbox environments. This enables testing, debugging, file processing, data analysis, web app prototyping, automation tasks, and more directly from the chat interface.

Availability: Available to all users.

Supported Languages

CodingFleet supports 20+ programming languages, including:

Python, JavaScript, TypeScript, Ruby, Go, Rust, C, C++, C#, Java, PHP, Perl, R, Kotlin, and others.

Additional languages may be added over time. If you need support for a specific language, contact us and we'll consider adding it.

How It Works

When Code Execution is enabled, the AI gains access to a secure, isolated sandbox environment.

Sandbox Lifecycle

Your sandbox remains available while you actively use Code Execution.

When the sandbox is inactive for several minutes, CodingFleet automatically saves it as a snapshot image and shuts it down to save resources. When you need it again, the sandbox reloads from the most recent snapshot.

You can change the snapshot image from the Sandbox Management page (requires login).

First-Time Setup

The first execution of a specific language or environment may take longer because required runtimes and dependencies may need to be installed. Subsequent executions are faster.

File Uploads

You can upload files directly into your sandbox environment.

Uploaded files are placed inside the /uploads directory and can be accessed by the AI during execution.

File upload support has been significantly improved and now supports hundreds of file types.

File Downloads

The AI can generate downloadable files and provide them back to you, including:

  • CSV files
  • JSON files
  • Reports
  • PDFs
  • Images
  • Charts and graphs
  • ZIP archives
  • Generated source code
  • Any other files created inside the sandbox

You can also download files and folders directly from the Sandbox Manager.

Web Application Hosting

You can instruct the AI to build and run web applications inside the sandbox.

The AI can:

  • Launch web servers
  • Host temporary applications
  • Provide public URLs
  • Test APIs
  • Validate endpoints
  • Perform automated testing

For persistent applications, instruct the AI to run services in the background so execution can continue while additional tasks are performed.

Hosting on External Providers

You can also provide a temporary, scoped API key for a web hosting provider or VM provider, such as DigitalOcean, Replit, or similar services.

With your instructions, the AI can use that temporary key to help deploy or host your code automatically.

For security, we recommend using temporary keys with limited permissions and deleting or revoking them after the task is complete.

Advanced Sandbox Capabilities

When Code Execution is enabled, the AI can use specialized tools to interact with the sandbox filesystem and environment.

Filesystem Tools

  • fs_read — Read files with line numbers.
  • fs_write — Create or overwrite files.
  • fs_edit — Modify existing files.
  • fs_glob — Find files using glob patterns.

Command Execution

The AI can execute shell commands directly inside the sandbox using Bash commands, Linux utilities, development tools, build systems, and package managers.

Database Access

The AI can execute SQL queries against databases running inside the sandbox. Supported databases include:

PostgreSQL, SQLite, MySQL, Redis, and MongoDB.

Sandbox Management

CodingFleet provides a dedicated Sandbox Management interface.

Once you sign up and login to your account, using Sandbox Management, you can:

  • Create sandboxes
  • Manage files
  • Access shell sessions
  • Download/Upload files and folders
  • Create snapshots
  • Restore snapshots
  • Change the snapshot image used for reloads
  • Build from Docker images
  • Terminate sandboxes when no longer needed

Projects Integration

Code Execution works with Projects.

You can organize chats, files, generated assets, and sandbox work inside Projects, allowing the AI to use that grouped context while working on larger tasks.

Supported Models

All current CodingFleet models support Code Execution.

Model availability may still depend on your subscription plan.

Performance Improvements

Recent improvements include:

  • Up to 25× faster sandbox file uploads
  • Improved filesystem handling
  • Faster file downloads
  • Better sandbox stability
  • More reliable long-running executions
  • Improved shell and process management
  • Better handling of generated files and folders

Current Limitations

  • Very long-running processes may be terminated automatically.
  • Sandbox environments are temporary and may reload from the latest snapshot after inactivity.
  • External hosting depends on the permissions and limits of the temporary API key you provide.
  • Some providers may require extra manual confirmation, billing setup, or account configuration outside CodingFleet.

Need Help?

If you encounter issues with Code Execution or need support for additional languages, runtimes, packages, or deployment workflows, please contact us.

   Last Updated: jun 2026