---
title: "E2B Integrations"
description: "Coding agents, agent frameworks, RL environments and evals that run on E2B sandboxes."
url: "https://e2b.dev/integrations"
---

# E2B Integrations

> Coding agents, agent frameworks, RL environments and evals that already run on E2B sandboxes. Any Python or JavaScript library runs inside an E2B sandbox, listed here or not.

Entries linking to `https://e2b.dev/integrations/<slug>.md` have an E2B setup guide in Markdown: a prompt for your coding agent, the install and code steps, what runs in the sandbox, and the docs link. The rest link to their docs or repository.

## Coding agents

- [OpenClaw](https://e2b.dev/integrations/openclaw.md): Host an always-on OpenClaw agent in a sandbox, with its gateway reachable from chat apps.
- [Hermes Agent](https://e2b.dev/integrations/hermes.md): Run Hermes Agent in a sandbox template, or keep it local and execute its terminal tools on E2B.
- [OpenCode](https://e2b.dev/integrations/opencode.md): Run OpenCode in a sandbox, or keep it local and point its workspaces at E2B with the @e2b/opencode plugin.
- [Claude Code](https://e2b.dev/integrations/claude-code.md): Run Claude Code in its own sandbox with a real filesystem, terminal, and git, so it can work unattended.
- [OpenAI Codex](https://e2b.dev/integrations/codex.md): Run OpenAI's Codex CLI headless in a sandbox and collect the diff when it's done.
- [Pi](https://e2b.dev/integrations/pi.md): Run the Pi coding agent in a sandbox, or keep Pi local and send its tool calls to E2B with @e2b/pi.
- [Gemini CLI](https://e2b.dev/integrations/gemini-cli.md): Run Google's Gemini CLI in a sandbox with full filesystem, terminal, and git access.
- [OpenHands](https://e2b.dev/integrations/openhands-canvas.md): Run OpenHands in a sandbox and open its web UI from your browser.
- [Prime Agent](https://e2b.dev/integrations/prime.md): Run Prime Intellect's self-improving Prime Agent from a prebuilt template.
- [QM](https://e2b.dev/integrations/qm.md): Deploy QM, Y Combinator's multiplayer agent harness, with an isolated sandbox for every person and room.
- [GitHub Copilot CLI](https://e2b.dev/integrations/copilot-cli.md): Run GitHub Copilot CLI in a sandbox with full filesystem, terminal, and git access.
- [Sandcastle](https://e2b.dev/integrations/sandcastle.md): Run Claude Code or Codex in parallel E2B sandboxes with Sandcastle and get the commits back in your local repo.
- [Kimi Code CLI](https://e2b.dev/integrations/kimi-code-cli.md): Run Moonshot AI's Kimi Code CLI in a sandbox with full filesystem, terminal, and git access.
- [Amp](https://docs.e2b.dev/agents/amp): Run Amp against a sandboxed copy of your repo, with full terminal and file access.
- [Antigravity CLI](https://docs.e2b.dev/agents/antigravity-cli): Run Google's Antigravity CLI in a sandbox with full filesystem, terminal, and git access.
- [Devin CLI](https://docs.e2b.dev/agents/devin): Run the Devin CLI inside a prebuilt sandbox template.
- [Factory Droid](https://docs.e2b.dev/agents/droid): Run Factory's Droid in a sandbox with full filesystem, terminal, and git access.
- [Grok Build](https://docs.e2b.dev/agents/grok): Run xAI's Grok Build from a ready-made sandbox template.
- [Kiro CLI](https://docs.e2b.dev/agents/kiro): Run the Kiro CLI in a sandbox with full filesystem, terminal, and git access.
- [Meta Muse Code](https://docs.e2b.dev/agents/muse): Use E2B as the execution backend for Meta's Muse Code CLI.

## Agent platforms

- [Paperclip](https://docs.paperclip.ing/reference/adapters/sandbox-providers/): Use E2B as the sandbox provider for agents you manage in Paperclip.
- [Agent Client Protocol (ACP)](https://docs.e2b.dev/agents/acp): Run goose, OpenCode, Claude Code, or any ACP agent in a sandbox and drive it from your code.
- [Claude Managed Agents](https://e2b.dev/integrations/claude-managed-agents.md): Let Claude Managed Agents run their tools in sandboxes you own, one per session.
- [Brainbase](https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/brainbase-universal-harness-bakeoff): Run Brainbase's universal agent harness on E2B and compare harnesses side by side.
- [Cursor cloud agents](https://docs.e2b.dev/agents/cursor): Run Cursor's self-hosted cloud agents on E2B dispatcher and worker sandboxes.
- [Devin Outposts](https://docs.e2b.dev/agents/devin-outposts): Give Devin Outposts its execution machines on E2B instead of running your own fleet.
- [Gemini Managed Agents](https://docs.e2b.dev/agents/gemini-managed-agents): Run the file and shell tools of a Gemini API managed agent in your own E2B sandbox.
- [OpenAI Agents API](https://docs.e2b.dev/agents/openai-agents-api): Pick E2B as the execution environment for OpenAI's Agents API, app-managed or webhook-managed.

## Agent frameworks & SDKs

- [CrewAI](https://e2b.dev/integrations/crewai.md): Give CrewAI agents E2B tools for shell commands, Python, and files with the crewai-tools[e2b] extra.
- [LangGraph](https://e2b.dev/integrations/langgraph.md): Run code from LangGraph plan-and-execute agents in an E2B sandbox.
- [Agno](https://e2b.dev/integrations/agno.md): Add Agno's built-in E2BTools so agents can run code and commands in a sandbox.
- [AgentScope](https://github.com/agentscope-ai/agentscope): Use an E2B workspace for AgentScope agents.
- [LangChain Deep Agents](https://e2b.dev/integrations/langchain-deepagents.md): Back LangChain Deep Agents with an E2B sandbox through the langchain-e2b package.
- [OpenAI Agents SDK](https://e2b.dev/integrations/openai-agents-sdk.md): Use the built-in E2B sandbox client in the OpenAI Agents SDK to give agents a real machine.
- [Hugging Face smolagents](https://github.com/huggingface/smolagents): Set executor_type="e2b" and smolagents runs every code action in a sandbox.
- [Mastra](https://e2b.dev/integrations/mastra.md): Give Mastra agents an E2B workspace or a full desktop with @mastra/e2b and @mastra/e2b-desktop.
- [Vercel AI SDK](https://e2b.dev/integrations/vercel-ai-sdk.md): Give Vercel AI SDK agents sandbox tools with @e2b/ai-sdk-sandbox. Experimental for AI SDK 7.
- [Haystack](https://haystack.deepset.ai/integrations/e2b): Give Haystack agents an E2B sandbox toolset with the e2b-haystack integration.
- [Google ADK](https://e2b.dev/integrations/google-adk.md): Use E2B as the code execution environment for Google ADK agents in Python.
- [Pydantic AI](https://e2b.dev/integrations/pydantic-ai.md): Run Pydantic AI agent tools in E2B through the pydantic-ai-harness E2B sandbox.
- [CAMEL](https://github.com/camel-ai/camel): Run CAMEL multi-agent code through its built-in E2B interpreter.
- [VoltAgent](https://github.com/VoltAgent/voltagent): Give VoltAgent agents an E2B sandbox with @voltagent/sandbox-e2b.
- [Flue](https://e2b.dev/integrations/flue.md): Add E2B to a Flue project with one command and run its sandbox agents on it.
- [Vercel Eve](https://docs.e2b.dev/agents/eve): Use E2B as the sandbox backend for Vercel's Eve agent framework. Experimental.
- [TanStack AI](https://tanstack.com/ai/latest/docs/sandbox/providers): Plug E2B into TanStack AI as a full sandbox or as a Code Mode isolate.
- [VibeKit](https://github.com/superagent-ai/vibekit): Run Claude Code, Codex, or Gemini CLI in E2B through VibeKit's agent sandbox SDK.
- [Rivet Sandbox Agent](https://rivet.dev/sandbox-agent/docs/deploy/e2b/): Deploy Rivet's Sandbox Agent server into E2B and drive any coding agent over HTTP.
- [CUGA](https://github.com/cuga-project/cuga-agent): Execute CUGA's generated code in E2B with its built-in executor.
- [Dynamiq](https://github.com/dynamiq-ai/dynamiq): Use E2B as a sandbox backend in the Dynamiq orchestration framework.
- [Inngest AgentKit](https://agentkit.inngest.com/integrations/e2b): Build an AgentKit coding agent whose tools run in E2B, orchestrated by Inngest.
- [Stirrup](https://github.com/ArtificialAnalysis/Stirrup): Run Artificial Analysis's lightweight Stirrup agents with E2B code execution.
- [ComputeSDK](https://github.com/computesdk/computesdk): Call E2B through ComputeSDK's provider-agnostic sandbox API.
- [Molecule](https://github.com/molecule-dev/molecule): E2B code-sandbox bond for Molecule's composable full-stack packages.
- [GAIA Agent](https://github.com/gaia-agent/gaia-agent): E2B execution provider for the GAIA-benchmark agent built on the AI SDK.
- [agentsh](https://github.com/canyonroad/agentsh-secure-sandbox): Wrap E2B sandboxes with agentsh's policy-enforced secure shell.
- [Fountain](https://github.com/managoat/managoat): Expose Claude Code, Codex, and Gemini CLI as an HTTP API, each in its own E2B sandbox.
- [agentproto](https://github.com/agentproto/ts): E2B sandbox provider for the agentproto TypeScript runtime.
- [Digger Open Agent SDK](https://github.com/diggerhq/open-agent-sdk): Use E2B as the pluggable sandbox in Digger's Open Agent SDK.
- [Sandbar](https://github.com/beachside-compute/beachside): E2B provider for Sandbar's one-SDK-for-every-sandbox CLI.
- [Amends](https://github.com/amends-dev/amends): E2B sandbox adapter for Amends.
- [PAREL](https://github.com/parall-hq/parel-oss): E2B sandbox plugin for PAREL agents.
- [Tangle agent SDK](https://github.com/tangle-network/agent-sdk): E2B provider for the Tangle agent SDK.
- [ctx-zip](https://www.npmjs.com/package/ctx-zip): Run ctx-zip's code-mode tool compression on an E2B sandbox.
- [EdgeOne Pages Agent Toolkit](https://www.npmjs.com/package/@edgeone/pages-agent-toolkit): Use E2B command, file, and browser adapters in the EdgeOne Pages agent toolkit.
- [Mayura](https://mayurajs.com/integrations/e2b/): Give Mayura agents an E2B sandbox with @mayurajs/sandbox-e2b.

## Training & evals

- [DSPy RLMs](https://e2b.dev/integrations/dspy.md): Run DSPy recursive language model (RLM) code in E2B sandboxes.
- [Hugging Face Open-R1](https://github.com/huggingface/open-r1): Score code-generation rewards in E2B while reproducing DeepSeek-R1 with Open-R1.
- [Hugging Face TRL](https://e2b.dev/integrations/trl.md): Compute code rewards for TRL GRPO training in E2B sandboxes.
- [Harbor / Terminal-Bench](https://e2b.dev/integrations/harbor.md): Run Harbor and Terminal-Bench trials on E2B and scale evals and rollouts to thousands of parallel sandboxes.
- [Miles](https://miles.radixark.com/docs/user-guide/sandbox-providers): Use E2B as the sandbox provider for Miles RL post-training rollouts.
- [Hugging Face OpenEnv](https://e2b.dev/integrations/openenv.md): Run OpenEnv environments like Terminus on E2B and get a first reward without a GPU.
- [SkyRL](https://github.com/NovaSky-AI/SkyRL): Train with SkyRL on Harbor tasks running in E2B sandboxes.
- [NVIDIA NeMo Gym](https://docs.e2b.dev/agents/nemo-gym): Run NVIDIA NeMo Gym evaluation and training environments on E2B.
- [OpenEvals](https://github.com/langchain-ai/openevals): Execute and grade generated code in E2B with LangChain's OpenEvals evaluators.
- [EdgeBench / SForge](https://github.com/ByteDance-Seed/EdgeBench): Run ByteDance Seed's EdgeBench / SForge environments on the E2B backend.
- [Scale AgentEnv](https://github.com/scaleapi/agentenv-framework): Build RL environments with Scale's AgentEnv framework on an E2B sandbox provider.
- [NiceEval](https://github.com/NiceEval/NiceEval): Run NiceEval agent evals in E2B sandboxes.
- [Strands Env](https://github.com/open-rollout/strands-env): Use prebaked E2B environments for agentic RL with Strands Env.
- [SWE-bench runner](https://github.com/e2b-dev/swe-bench): Evaluate SWE-bench instances in E2B sandboxes with the reference runner.
- [CyberGym](https://github.com/e2b-dev/cybergym-e2b): Run the CyberGym security benchmark on E2B, unmodified.
- [Inspect AI](https://github.com/meridianlabs-ai/inspect_sandboxes): Run Inspect AI evals in E2B with the inspect-sandboxes provider.
- [OSWorld](https://github.com/e2b-dev/osworld-e2b): Run the OSWorld computer-use benchmark on E2B Desktop sandboxes.
- [NVIDIA SkillEvaluator](https://docs.nvidia.com/skills/skillevaluator): Evaluate agent skills with NVIDIA SkillEvaluator on E2B.
- [Veris](https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/veris-e2b-js): Simulate and test agents with Veris, with agent code running in E2B.

## Browser & computer use

- [Playwright](https://docs.e2b.dev/use-cases/remote-browser): Run Playwright inside a sandbox, or connect it to a remote browser from one.
- [Cua](https://github.com/trycua/cua): Run Cua computer-use agents on E2B with --on e2b.
- [CopilotKit OpenFuse](https://github.com/CopilotKit/openfuse): CopilotKit's personal agent uses an E2B Desktop for its browser and terminal.
- [Notte](https://github.com/nottelabs/notte): Run Notte web agents with their code executing in E2B.
- [Kernel](https://e2b.dev/integrations/kernel.md): Pair an E2B sandbox with a Kernel cloud browser over CDP.
- [Claude browser and computer-use toolsets](https://docs.e2b.dev/agents/claude-toolsets): Back Claude's browser and computer-use toolsets with E2B drivers in Python or TypeScript.
- [Anchor Browser](https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/anchor-cloud-browser-js): Drive an Anchor cloud browser from code running in E2B.
- [OpenAI computer use](https://docs.e2b.dev/agents/openai-computer-use): Give OpenAI's computer-use model an E2B Desktop to see and control.

## Tools & MCP

- [Firecrawl](https://docs.e2b.dev/mcp-gateway/available-servers): Scrape the web with Firecrawl, then analyze the results in a sandbox. Also in the MCP Gateway.
- [Composio](https://docs.composio.dev/toolkits/e2b): Call E2B from Composio's toolkit catalog alongside 1,000+ other tools.
- [MCP Gateway](https://e2b.dev/integrations/mcp.md): Turn on 200+ MCP servers inside any sandbox and connect your agent to them through one gateway.
- [Arcade](https://docs.arcade.dev/en/resources/integrations/development/e2b): Use Arcade's E2B toolkit so agents can run code through Arcade-authorized tools.

## Workflows & automation

- [n8n](https://e2b.dev/integrations/n8n.md): Run code and commands in E2B sandboxes from n8n workflows with the official E2B node.
- [Dify](https://marketplace.dify.ai/plugin/langgenius/e2b): Add the E2B plugin from the Dify Marketplace to run code in Dify workflows.
- [Convex](https://github.com/e2b-dev/e2b-convex): Create and manage E2B sandboxes from Convex actions with @e2b/convex.
- [Pipedream](https://pipedream.com/apps/e2b/actions/run-code): Run code in E2B from Pipedream workflows with the prebuilt action.
- [Hatchet](https://github.com/hatchet-dev/hatchet): Run Hatchet durable tasks in E2B sandboxes through Hatchet Cloud's sandbox operator.
- [Better E2B Sandbox for Dify](https://marketplace.dify.ai/plugin/lysonober/better-e2b-sandbox): A community Dify plugin with extra E2B sandbox controls.
- [Make](https://www.make.com/integrations/e2b): Run code in E2B as a step in Make scenarios.
- [Relevance AI](https://relevanceai.com/integrations/e2b): Give Relevance AI agents E2B code execution.

## Model providers

- [Anthropic models](https://docs.e2b.dev/quickstart/connect-llms): Connect Claude tool calls to E2B code execution.
- [Fireworks AI](https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/fireworks-code-interpreter-python): Give Fireworks-hosted open models a code interpreter on E2B.
- [Google Gemini models](https://docs.e2b.dev/agents/google-adk): Connect Gemini to E2B code execution through Google ADK.
- [Groq](https://docs.e2b.dev/quickstart/connect-llms): Pair fast Groq inference with E2B code execution. Groq's own docs use E2B for server-side code.
- [IBM watsonx.ai](https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/watsonx-ai-code-interpreter-js): Add a code interpreter to IBM watsonx.ai models with E2B.
- [Mistral / Codestral](https://docs.e2b.dev/quickstart/connect-llms): Let Codestral and other Mistral models run the code they write in E2B.
- [OpenAI models](https://docs.e2b.dev/quickstart/connect-llms): Connect OpenAI model tool calls to E2B code execution.
- [Together AI](https://github.com/e2b-dev/e2b-cookbook/tree/main/examples/together-ai-code-interpreter-js): Run code from Together AI-hosted open models in an E2B code interpreter.

## Infrastructure & dev tools

- [Claude Code Templates](https://github.com/davila7/claude-code-templates): Install E2B components for Claude Code from the claude-code-templates registry.
- [NVIDIA OpenShell](https://github.com/e2b-dev/openshell-e2b): Run NVIDIA OpenShell agent sessions on E2B.
- [Tailcat](https://github.com/tailscale/tailcat): Reach private services from a sandbox over Tailscale's data plane with Tailcat.
- [Crabbox](https://docs.e2b.dev/agents/crabbox): Warm an E2B box, sync your diff, and run the test suite remotely with Crabbox.
- [Claude Code Kit](https://github.com/padak/claude-code-kit): A Claude Code plugin kit with E2B sandbox commands.
- [Herdr](https://github.com/e2b-dev/herdr-e2b-sandbox): Mirror a git worktree into E2B: one box, or a branch-per-agent fleet with a TUI dashboard.
- [E2B Actions Runner for VS Code](https://marketplace.visualstudio.com/items?itemName=iotserver24.e2b-actions-runner): Run code from VS Code in E2B sandboxes.
- [Archil](https://docs.e2b.dev/storage/archil): Mount Archil storage into sandboxes so files persist across runs.
- [E2B Sandbox Explorer for VS Code](https://marketplace.visualstudio.com/items?itemName=bhavaniravi.e2b-sandbox-explorer): Browse and manage E2B sandboxes from the VS Code sidebar.
- [Machine Payments Protocol](https://docs.e2b.dev/use-cases/agent-payments): Let agents pay for sandboxes per request over MPP, with no E2B account or API key.
- [Raycast](https://www.raycast.com/e2b/e2b): Start and manage sandboxes from Raycast.
- [Stripe Projects](https://docs.stripe.com/projects): Provision E2B sandboxes and an API key from the Stripe CLI with Stripe Projects.
- [Vercel Connect](https://vercel.com/connect/e2b): Connect E2B to your Vercel projects through Vercel Connect.

## Built on E2B

- [Open Lovable](https://github.com/firecrawl/open-lovable)
- [ClaraVerse](https://github.com/claraverse-space/ClaraVerse)
- [MIRIX](https://github.com/Mirix-AI/MIRIX)
- [AxisAgentic](https://github.com/XYZ-AI-Lab/AxisAgentic)
- [MemoRizz](https://github.com/RichmondAlake/memorizz)
- [MM-Agent](https://github.com/usail-hkust/LLM-MM-Agent)
- [NVIDIA QeRL](https://github.com/NVlabs/qerl)
- [ABC-GRPO](https://github.com/chi2liu/ABC-GRPO)
- [MindRoom](https://github.com/mindroom-ai/mindroom)
- [Agentica](https://github.com/shibing624/agentica)
- [Open-RS](https://github.com/knoveleng/open-rs)
- [AutoGEO](https://github.com/cxcscmu/AutoGEO)
- [Agentor](https://github.com/CelestoAI/agentor)
- [OS1](https://github.com/nokodo-labs/os1)
- [AOrchestra](https://github.com/FoundationAgents/AOrchestra)
- [Weaver](https://github.com/skygazer42/Weaver)
- [Research Claw](https://github.com/nanoAgentTeam/research-claw)
- [AgentMatrix](https://agentmatrix.pro/docs/en/environment-specs)
- [AnyCode](https://anycode.quantlix.com/latest/guides/sandbox-providers/)
- [OpenAgentCore](https://openagentcore.dev/docs/sandbox-provider)

## Docs

- [E2B documentation](https://docs.e2b.dev)
