# Docs - **Introduction** - [Daytona Documentation](/docs/quickstart): Start managing your sandboxes with Daytona. - [Authentication](/docs/authentication): Authenticate the Daytona SDKs, CLI, and API to create sandboxes. - [Sandboxes](/docs/sandboxes): Isolated virtual machines for autonomous agents, code execution, and model training. - **Sandbox** - [Linux Sandboxes](/docs/linux-sandboxes): Linux virtual machine sandboxes, their capabilities, use cases, and architecture. - [Windows Sandboxes](/docs/windows-sandboxes): Windows virtual machine sandboxes and their capabilities. - [macOS Sandboxes](/docs/macos-sandboxes): macOS virtual machine sandboxes, their capabilities, use cases, and architecture. - [GPU Sandboxes](/docs/gpu-sandboxes): GPU-accelerated sandboxes and their capabilities. - **Concepts** - [Isolation](/docs/isolation): Sandbox runtime, network, and project isolation. - [Persistence](/docs/persistence): Keeping sandbox state beyond the life of a sandbox. - [Scale](/docs/scale): Sandbox scalability and concurrency. - [Preemption](/docs/preemption): Discounted spot sandboxes that Daytona can reclaim when it needs capacity. - **Configuration** - [Snapshots](/docs/snapshots): Captures of the memory and disk of a running sandbox that new sandboxes resume from. - [Images](/docs/images): Register, build, and capture the images that sandboxes boot from. - [Registries](/docs/registries): Store credentials for private container registries and pull private images into sandboxes. - [Volumes](/docs/volumes): Persistent storage that sandboxes attach at creation and that outlives any single sandbox. - [Regions](/docs/regions): Geographic locations where sandboxes run. - [Network Limits](/docs/network-limits): Control network access to your sandboxes. - **Toolbox** - [File System Operations](/docs/file-system-operations): Interact with the file system in your sandboxes. - [Git Operations](/docs/git-operations): Interact with Git repositories in your sandboxes. - [Language Server Protocol](/docs/language-server-protocol): Interact with language server protocol servers in your sandboxes. - [Code Execution](/docs/code-execution): Run Python, TypeScript, and JavaScript code in your sandboxes through language kernels. - [Process Execution](/docs/process-execution): Run shell commands and manage long-running processes in your sandboxes. - [Pseudo Terminal (PTY)](/docs/pty): Interact with pseudo terminal sessions in your sandboxes. - [Log Streaming](/docs/log-streaming): Stream logs as they are being produced, while the process is still running. - [Daytona MCP Server](/docs/mcp): Interact with Daytona sandboxes using the Model Context Protocol. - [Computer Use](/docs/computer-use): Programmatic control of desktop environments within Daytona sandboxes. - [Docker](/docs/docker): Run Docker Engine inside Daytona sandboxes. - **Access** - [Web Terminal](/docs/web-terminal): Interact with your sandboxes using a browser-based web terminal. - [SSH Access](/docs/ssh-access): Access your sandboxes over SSH with a sandbox access token or a project API key. - [VNC Access](/docs/vnc-access): Interact with your sandbox's graphical desktop environment directly in the browser. - [VPN Connections](/docs/vpn-connections): Connect your sandboxes to private networks. - [Preview URLs](/docs/preview): Access services running in your sandboxes through preview URLs. - [Playground](/docs/playground): Write and run TypeScript against the Daytona SDK in your browser. - **Security** - [Secrets](/docs/secrets): Inject project secrets into sandbox network traffic without exposing the plaintext to sandbox code. - [Audit Logs](/docs/audit-logs): View and monitor all user actions across your Daytona organization. - [OpenTelemetry](/docs/opentelemetry): Export logs, traces, and metrics from sandboxes and from the Daytona SDKs over OpenTelemetry. - Typescript sdk - [TypeScript SDK Reference](/docs/typescript-sdk): Interact with Daytona Sandboxes using the TypeScript SDK - [Daytona](/docs/typescript-sdk/daytona) - [Sandbox](/docs/typescript-sdk/sandbox) - [FileSystem](/docs/typescript-sdk/file-system) - [Git](/docs/typescript-sdk/git) - [Process](/docs/typescript-sdk/process) - [Interpreter](/docs/typescript-sdk/interpreter) - [ComputerUse](/docs/typescript-sdk/computer-use) - [Lsp](/docs/typescript-sdk/lsp) - [Ports](/docs/typescript-sdk/ports) - [Docker](/docs/typescript-sdk/docker) - [Image](/docs/typescript-sdk/image) - [Snapshot](/docs/typescript-sdk/snapshot) - [Volume](/docs/typescript-sdk/volume) - [Reservations](/docs/typescript-sdk/reservations) - [MacModels](/docs/typescript-sdk/mac-models) - [Secret](/docs/typescript-sdk/secret) - [Shared errors](/docs/typescript-sdk/shared-errors) - [Errors](/docs/typescript-sdk/errors) - Python sdk - [Python SDK Reference](/docs/python-sdk): Interact with Daytona Sandboxes using the Python SDK - [Daytona](/docs/python-sdk/daytona) - [Sandbox](/docs/python-sdk/sandbox) - [FileSystem](/docs/python-sdk/file-system) - [Git](/docs/python-sdk/git) - [Process](/docs/python-sdk/process) - [Interpreter](/docs/python-sdk/interpreter) - [ComputerUse](/docs/python-sdk/computer-use) - [Lsp](/docs/python-sdk/lsp) - [Ports](/docs/python-sdk/ports) - [Docker](/docs/python-sdk/docker) - [Image](/docs/python-sdk/image) - [Snapshot](/docs/python-sdk/snapshot) - [Volume](/docs/python-sdk/volume) - [Reservations](/docs/python-sdk/reservations) - [MacModels](/docs/python-sdk/mac-models) - [Secret](/docs/python-sdk/secret) - [Shared errors](/docs/python-sdk/shared-errors) - [Errors](/docs/python-sdk/errors) - Api - [Daytona API Reference](/docs/api): A reference of supported operations using the Daytona API. - Daytona API - [Daytona API](/docs/api/daytona) - [Sandboxes](/docs/api/daytona/sandboxes) - [Images](/docs/api/daytona/images) - [Migration](/docs/api/daytona/migration) - [Snapshots](/docs/api/daytona/snapshots) - [Volumes](/docs/api/daytona/volumes) - [Secrets](/docs/api/daytona/secrets) - [Registries](/docs/api/daytona/registries) - [Reservations](/docs/api/daytona/reservations) - [Previews](/docs/api/daytona/previews) - [Regions](/docs/api/daytona/regions) - [Catalog](/docs/api/daytona/catalog) - [Usage](/docs/api/daytona/usage) - [Users](/docs/api/daytona/users) - [Organizations](/docs/api/daytona/organizations) - [Projects](/docs/api/daytona/projects) - [Roles](/docs/api/daytona/roles) - [API Keys](/docs/api/daytona/api-keys) - [Service Accounts](/docs/api/daytona/service-accounts) - [Billing](/docs/api/daytona/billing) - [Audit](/docs/api/daytona/audit) - [System](/docs/api/daytona/system) - [Auth](/docs/api/daytona/auth) - Toolbox API - [Toolbox API](/docs/api/toolbox) - [Info](/docs/api/toolbox/info) - [Env](/docs/api/toolbox/env) - [Processes](/docs/api/toolbox/processes) - [Sessions](/docs/api/toolbox/sessions) - [Files](/docs/api/toolbox/files) - [Git](/docs/api/toolbox/git) - [LSP](/docs/api/toolbox/lsp) - [Ports](/docs/api/toolbox/ports) - [System](/docs/api/toolbox/system) - [Volumes](/docs/api/toolbox/volumes) - [Docker](/docs/api/toolbox/docker) - [Computer Use](/docs/api/toolbox/computer-use) - Cli - [CLI](/docs/cli): A reference of supported operations using the Daytona CLI. - [Login and context](/docs/cli/login-and-context): Reference for the daytona login, logout, profile, and whoami commands. - [Sandboxes and resources](/docs/cli/sandboxes-and-resources): Reference for the daytona cpu-type, image, mac-model, region, reservation, sandbox, snapshot, and volume commands. - [Organization management](/docs/cli/organization-management): Reference for the daytona audit, billing, invite, key, member, org, org-key, project, registry, role, secret, and service-account commands. - [Other](/docs/cli/other): Reference for the daytona completion, config, docs, and version commands. - [Architecture](/docs/architecture): Daytona platform architecture overview. - [Organizations](/docs/organizations): Manage your organizations and their resources. - [Projects](/docs/projects): Create and manage projects within your organization. - [API Keys](/docs/api-keys): Create and manage API keys for your projects. - [SSO](/docs/sso): Configure single sign-on and directory sync for your organization through the Admin Portal. - [Limits](/docs/limits): Overview of sandbox resource limits, organization limits, and API rate limits. - [Billing](/docs/billing): Overview of your organization's billing, wallet, and usage details. - Guides - [Guides](/docs/guides): Guides for using Daytona. - Agentkit - [Build Coding Agent Using AgentKit and Daytona](/docs/guides/agentkit/inngest-agentkit-coding-agent): Step-by-step guide to building an autonomous coding agent using AgentKit framework and Daytona sandboxes. - OpenCode - [OpenCode](/docs/guides/opencode): Guides for running OpenCode with Daytona sandboxes - [Run OpenCode with the Daytona Plugin](/docs/guides/opencode/opencode-plugin): Use the Daytona OpenCode plugin to run each OpenCode session in a Daytona sandbox, with automatic git sync to local branches. - [Build a Coding Agent Using OpenCode SDK and Daytona](/docs/guides/opencode/opencode-sdk-agent): Step-by-step guide to building an autonomous coding agent using OpenCode and Daytona sandboxes. - [Build a Coding Agent Using OpenCode Web and Daytona](/docs/guides/opencode/opencode-web-agent): Step-by-step guide to running OpenCode's web interface, an open source coding agent, inside Daytona sandboxes. - Pi - [Run Pi with the Daytona Extension](/docs/guides/pi/pi-extension): Use the Daytona Pi extension to run every Pi tool call inside a remote Daytona sandbox, with each session synced to its own GitHub branch. - Langchain - [Analyze Data With LangChain AI Agent](/docs/guides/langchain/langchain-data-analysis): Build LangChain agents that perform secure data analysis using Daytona's isolated sandbox environment. - Langgraph - [Build a Plan-and-Execute Data Agent With LangGraph and Daytona](/docs/guides/langgraph/langgraph-plan-and-execute-data-agent): Wire a six-node LangGraph state machine that runs an end-to-end ETL plus analytical-SQL workflow in an isolated Daytona sandbox, with structured planning and per-step retry recovery. - Mastra - [Use Mastra Coding Agent with Daytona](/docs/guides/mastra/mastra-coding-agent): Learn how to integrate Daytona sandboxes with the Mastra coding agent for secure, isolated AI-powered development. - Claude - [Claude](/docs/guides/claude): Guides for using Claude with Daytona - [Build a Two-Agent Coding System with Claude and Daytona](/docs/guides/claude/claude-agent-sdk-connect-service-sandbox): Step-by-step guide to building a dual-agent coding system using Claude Agent SDK and Daytona sandboxes. - [Build a Coding Agent Using Claude Agent SDK and Daytona](/docs/guides/claude/claude-agent-sdk-interactive-terminal-sandbox): Step-by-step guide to building an autonomous coding agent using Claude Code and Daytona sandboxes. - [Run Claude Code in a Daytona Sandbox via CLI](/docs/guides/claude/claude-code-run-cli-sandbox): Learn how to run Claude Code inside a Daytona sandbox using the Daytona CLI for secure and isolated task execution. - [Running Claude Code with Daytona](/docs/guides/claude/claude-code-run-tasks-stream-logs-sandbox): Step-by-step guide to running Claude Code with Daytona sandboxes. - [Let Claude Draw on a Daytona Sandbox Desktop](/docs/guides/claude/claude-draws-daytona-sandbox): Run Anthropic's computer toolset against a Daytona sandbox desktop, where Claude opens a sketchpad in Chromium and draws with the mouse. - [Run Claude Managed Agents on Daytona](/docs/guides/claude/claude-managed-agents): Run Claude Managed Agents inside your own Daytona sandboxes as a self-hosted environment. - [Analyze Data with AI](/docs/guides/claude/data-analysis-with-ai): Use Daytona to run AI-generated code for data analysis and visualization. - Brainbase - [Run Agents on Daytona with Brainbase's Universal Harness API](/docs/guides/brainbase/universal-harness-api): Use Brainbase's Universal Harness API to run any agent harness in an isolated Daytona sandbox with a single API call. - Gemini - [Run Gemini CLI Headlessly in Daytona and Stream Its Output](/docs/guides/gemini/gemini-cli-run-tasks-stream-logs-sandbox): Step-by-step guide to running the Gemini CLI as a headless coding agent inside Daytona sandboxes and streaming its task output back to the caller. - Devin - [Devin](/docs/guides/devin): Guides for running Devin with Daytona - [Run the Devin CLI in a Daytona Sandbox](/docs/guides/devin/devin-cli-run-tasks-stream-logs-sandbox): Step-by-step guide to running Cognition's Devin CLI as a coding agent inside Daytona sandboxes and returning its task output to the caller. - [Run Devin Outposts on Daytona](/docs/guides/devin/devin-outposts): Serve Devin Outposts sessions on Daytona sandboxes that you control. - Cursor - [Cursor](/docs/guides/cursor): Guides for running Cursor Cloud Agents with Daytona - [Run Cursor Self-Hosted Machines on Daytona](/docs/guides/cursor/cursor-self-hosted-machines): Run Cursor Cloud Agent workers in isolated Daytona sandboxes. - Kiro - [Run AWS Kiro's CLI in a Daytona Sandbox and Stream Its Output](/docs/guides/kiro/kiro-cli-run-tasks-stream-logs-sandbox): Step-by-step guide to running AWS Kiro's CLI as a coding agent inside Daytona sandboxes and streaming its task output back to the caller. - Kimi - [Run the Kimi CLI in a Daytona Sandbox](/docs/guides/kimi/kimi-cli-run-tasks-stream-logs-sandbox): Step-by-step guide to running Moonshot AI's Kimi CLI as a headless coding agent inside Daytona sandboxes and streaming its task output back to the caller. - OpenClaw - [OpenClaw](/docs/guides/openclaw): Guides for running OpenClaw with Daytona - [Run OpenClaw in a Daytona Sandbox via CLI](/docs/guides/openclaw/openclaw-run-secure-sandbox): Learn how to run OpenClaw inside a Daytona sandbox using the Daytona CLI for secure, isolated, and always-on AI assistant with Telegram and WhatsApp. - [Run OpenClaw in a Daytona Sandbox via SDK](/docs/guides/openclaw/openclaw-sdk-sandbox): Learn how to run OpenClaw inside a Daytona sandbox using the Daytona SDK — create a sandbox programmatically and access the dashboard via a preview URL. - Vercel - [Build a Multi-Language Benchmark Agent with the Vercel AI SDK and Daytona](/docs/guides/vercel/vercel-ai-sdk-multi-language-benchmark-agent-sandbox): Use the Vercel AI SDK's ToolLoopAgent with five sandbox tools to build a benchmarking agent that implements algorithms in Python and TypeScript, times them, plots the comparison with matplotlib, and downloads both the chart and a markdown findings report to your local filesystem - all running inside a Daytona sandbox. - OpenAI - [OpenAI](/docs/guides/openai): Guides for using OpenAI tools and SDKs with Daytona - [Build a Coding Agent Using Codex SDK and Daytona](/docs/guides/openai/codex-sdk-interactive-terminal-sandbox): Step-by-step guide to building an autonomous coding agent using OpenAI Codex and Daytona sandboxes. - [Run OpenAI Agents API Sessions in Daytona Sandboxes](/docs/guides/openai/openai-agents-api-self-hosted-sandbox): Connect OpenAI's hosted Agents API to a self-hosted Daytona sandbox — run the Codex executor, stream results, and manage the sandbox lifecycle. - [Using the OpenAI Agents SDK with Daytona Sandboxes](/docs/guides/openai/openai-agents-sdk-with-sandboxes): Core patterns for using the OpenAI Agents SDK with Daytona sandboxes: shell access, handoffs, memory, and human-in-the-loop workflows. - [Use OpenAI Computer-Use Helpers with Daytona Desktops](/docs/guides/openai/openai-computer-use-helpers): Let an OpenAI model operate a browser on a Daytona desktop with the ComputerToolset helpers in the OpenAI Python SDK. - [Text-to-SQL Agent with the OpenAI Agents SDK and Daytona](/docs/guides/openai/text-to-sql-agent-openai-agents-sdk): Build a conversational agent that translates natural-language questions into SQL queries against real NASA spending data. - Fireworks - [Run a Fireworks Coding Agent with BYOK Secrets](/docs/guides/fireworks/fireworks-coding-agent): Run coding agents on Fireworks AI open models inside Daytona sandboxes, protect the API key with host-scoped secrets, and export trajectories as fine-tuning data. - Amp - [Build a Coding Agent Using Amp Code and Daytona](/docs/guides/amp/amp-sdk-coding-agent): Step-by-step guide to building an autonomous coding agent using Amp Code CLI and Daytona sandboxes. - Google - [Generate Verified Code With Google ADK Agent](/docs/guides/google/google-adk-code-generator): Build Google ADK agents that generate and test code using Daytona's isolated sandbox environment. - Letta code - [Build a Coding Agent Using Letta Code and Daytona](/docs/guides/letta-code/letta-code-agent): Step-by-step guide to running Letta Code, a stateful coding agent with persistent memory, inside Daytona sandboxes. - Ag2 - [Fix Bugs Automatically With AG2 and Daytona](/docs/guides/ag2/ag2-bug-fixer-agent): Build a multi-agent bug fixer that analyzes broken code, proposes fixes, and verifies them in isolated Daytona sandboxes. - Flue - [Build an Autonomous Bug-Fix Agent with Flue and Daytona](/docs/guides/flue/flue-autonomous-bug-fix-agent): Use Flue's agent harness and a Daytona sandbox to reproduce a GitHub issue with a failing test, fix it, and open a real pull request, end to end. - Copilotkit - [Build a Generative-UI Coding Agent with CopilotKit and Daytona](/docs/guides/copilotkit/copilotkit-generative-ui-coding-agent-sandbox): Wire a CopilotKit Built-in Agent to a Daytona sandbox with full shell and filesystem access, where every tool call streams into the chat as generative UI and any hosted process embeds as a live iframe. - Convex - [Build an AI App Builder with Convex and Daytona](/docs/guides/convex/convex-ai-app-builder-sandbox): Build a full-stack AI app builder on the Daytona Convex component — an LLM writes an app, a Daytona sandbox runs it, and Convex streams every build step to the browser, ending with the running app embedded as a live iframe. - Model serving - [Serve LLMs on GPU Sandboxes with SGLang](/docs/guides/model-serving/sglang-gpu-model-serving): Serve OpenAI's gpt-oss reasoning model on a Daytona GPU sandbox with SGLang, and put it to work classifying documents with guaranteed-structure output. - [Serve LLMs on GPU Sandboxes with vLLM](/docs/guides/model-serving/vllm-gpu-model-serving): Spin up an OpenAI-compatible vLLM endpoint on a Daytona GPU sandbox and query it from anywhere. - Reinforcement learning - [Run an RL Coding Environment at Scale with HUD and Daytona](/docs/guides/reinforcement-learning/hud-rl-cookbook): Define a coding environment with HUD, scale it to 256 parallel rollouts on Daytona sandboxes, and train a model on the graded rollouts. - [Multi-Turn RL Training with OpenEnv and Daytona](/docs/guides/reinforcement-learning/openenv-finqa): Use parallel Daytona sandboxes with OpenEnv's FinQA environment to train Qwen3-14B on financial question answering via GRPO. - [Train LLMs with Reinforcement Learning Using TRL and Daytona](/docs/guides/reinforcement-learning/trl-grpo-training): Use parallel Daytona sandboxes to train code-generating LLMs with TRL's GRPO trainer. - [Scaling Tool Execution in RL Training with veRL and Daytona](/docs/guides/reinforcement-learning/verl-retool-daytona): Run veRL's ReTool recipe on Daytona sandboxes, configure the Daytona backend, and benchmark tool-calling throughput. - Recursive Language Model - [Recursive Language Model](/docs/guides/rlm): Guides for building and running Recursive Language Models (RLMs) with Daytona - [Run DSPy RLMs on Daytona](/docs/guides/rlm/dspy-rlms): Use DSPy's RLM module with a Daytona-backed interpreter to safely execute LLM-generated Python code in an isolated cloud sandbox. - [Build deep Recursive Language Models](/docs/guides/rlm/recursive-language-models): Implement recursive language model agents where each agent runs in its own isolated Daytona sandbox. ## Site - [About](https://www.daytona.com/about.md) - [Startup program](https://www.daytona.com/startups.md) - [Research grants](https://www.daytona.com/grants.md) - [Compute conference](https://www.daytona.com/compute.md) - [Pricing](https://www.daytona.com/pricing.md) - [Changelog](https://www.daytona.com/changelog.md)