Agent frameworks & infrastructure

Agent sandboxes & runtimes

1,721 agents · most starred first

Agent sandboxes & runtimes, page 16 of 18

xagent-ai

a powerful agent os with great compatibility

@ai-sdk/harness-opencode

The OpenCode harness connects `HarnessAgent` to OpenCode through a sandboxed bridge.

safe-lab-agents

Safely run AI agents in Docker to control scientific experiments via MCP

@stackbone/sdk

Official TypeScript SDK for Stackbone — the marketplace and managed runtime for containerized AI agents.

@livx.cc/agentx

Edge-native AI agent runtime — drives a virtual filesystem via any LLM (ai.libx.js). Same bytes run in node, browser, or edge.

agentic-runtime

Agentic Platform local execution runtime and CLI

@agentplatform/agentic-runtime

Agentic Platform local execution runtime and CLI

autowonder

AutoWonder local runtime — execute AI agent dispatch packages on your machine

stuntdouble-sandbox-cli

Secure AI Agent Sandbox with Docker & eBPF interceptors

feral-agent

Feral Agent — proactive, portable AI agent with a native security sandbox

@absolutejs/agent-runtime

Durable provider-neutral AI agent runs, steps, checkpoints, budgets, leases, timers, effects, and handoffs.

@tinysandbox/tinysandbox

Node.js bindings for the tinysandbox in-process AI agent sandbox

noumen

Programmatic AI agent runtime with pluggable providers and sandboxed virtual infrastructure

@hyperlight-dev/hyperagent

Interactive AI agent using GitHub Copilot SDK with Hyperlight sandboxed JS execution

skein-js

An open-source, self-hosted server runtime for LangGraph.js agents in TypeScript. A drop-in replacement for the LangGraph CLI (dev/up/build/dockerfile).

@workflow/world-vercel

Vercel platform World implementation for Workflow SDK

@agent-browser/sandbox

Helpers for running agent-browser inside sandbox runtimes

mycomesh-provider

Docker-backed MycoMesh Provider launcher with the Consumer CLI compatibility entry point.

@flower-f/afk-node

Node.js runtime adapters, workspace management, config, auth, and plugin support for afk-dev.

@skydiveai/pi-server
@provos/ironcurtain

Secure agent runtime with trusted process mediation

@ai-sdk/sandbox-just-bash

_This package is **experimental**._

@e2b/eve-sandbox

E2B sandbox backend for Vercel's Eve agent framework

@rlaope/agentbox

The AI Agent Framework — sandboxed multi-agent orchestration, high-throughput stateful agents, modular agent session manager & runtime for pi / codex / claude

@soddong/agentic-runtime

Agentic Platform local execution runtime and CLI

create-snowline-agents

Snowline Portable Agentic OS - Companion layer for AI agents

@octos-org/octos

Rust-native, API-first Agentic OS server (octos serve + bundled skills). Installs the prebuilt release bundle.

@aihu/compiler-native-darwin-x64

aihu compiler native addon (napi) — darwin-x64 platform binary.

@aihu/compiler-native-linux-arm64-gnu

aihu compiler native addon (napi) — linux-arm64-gnu platform binary.

toolweave

Typed code-mode tool orchestration for LLM agents: tools become TypeScript declarations, generated programs are type-checked (routeable checker backends), repaired via diagnostics, and executed in a QuickJS sandbox.

3tears-agent-workspace

workspace primitives for 3tears agents: workspace entities, sandbox, format handlers, tools

agent-runtime-cockpit

ARC — Agent Runtime Cockpit: Python daemon, adapters, context retrieval

agent-saga

Never worry about your AI agent breaking production again. Kill the process mid-flight — the wire transfer is still safely refunded, your database stays clean, and your users never lose trust. agent-saga is the 1-line bulletproof safety OS that turns unpredictable LLMs into crash-proof, enterprise-grade autonomous engines.

agent-vm-agent-portal-sdk

Portable Python contracts and clients for the agent-vm Tool Portal.

agent-vm-hermes-adapter

Partial managed Hermes profile adapter for agent-vm.

agentbox-rl

Docker sandboxes, tools, rewards and trajectories for RL post-training of LLM agents (TRL, verl, OpenRLHF, SkyRL, ART)

agentdeploy

Zero-boilerplate deployment for AI agent frameworks — LangGraph, CrewAI, OpenAI Agents SDK, and custom agents

agentic-compute-mcp-codelad1304

A secure, monetized Python sandbox for data optimization and Matplotlib rendering via x402.

agentkit-runtime

Agent Runtime SDK for igentbase — build sandboxed AI agents with one import

agentnest

A secure, self-hosted runtime for isolated AI agent execution

arbitrage-feather

Feather durable execution platform — Python SDK

arbtrage-feather-sdk

Feather durable execution platform — Python SDK

aws_sdk_bedrock_agentcore

Python client for Amazon Bedrock AgentCore

aws_sdk_bedrock_agentcore_control

Python client for Amazon Bedrock AgentCore Control

cocoonstack-sandbox-langchain

LangChain tools backed by cocoon microVM sandboxes

cocoonstack-sandbox-openai

OpenAI Agents SDK sandbox provider backed by cocoon microVMs

deepagents-opensandbox-backend

Deep Agents sandbox backend for a locally-run OpenSandbox server

inis-openai-agents

inis.run sandbox provider for the OpenAI Agents SDK (beta)

inis-pydantic-ai

Pydantic AI toolset for inis.run sandboxes

inis-strands-agents

inis.run sandbox provider for Strands Agents

langchain-baponi

Baponi sandbox provider for LangChain Deep Agents

langchain-inis

inis.run sandbox backend for LangChain Deep Agents, plus a LangGraph recipe

langchain-lightning

Lightning AI sandbox integration for Deep Agents

langgraph-runtime-pg

Self-hosted LangSmith Deployments alternative: MIT Postgres+Redis runtime (edition=pg) for stock langgraph-api

llm-cage

Cage: run local-LLM components in a network-isolated OS process, hardware access intact, with a friction-free call interface back to the rest of your app.

openclaw-agent-scheduler-sidecar-liquid

Hardware-aware scheduler sidecar for OpenClaw

sovereign-agentops-community

Runtime-agnostic governed agent execution — Community Edition

sprites-openai-agents

Sprites sandbox provider for the OpenAI Agents SDK

tb-agent

Local execution daemon for Tether Brain agent installations

urirun-llm-runtime

Canonical LLM-facing URI process runtime — spec, executor, CI gates

CorvinOS

CorvinOS is a secure, self-hosted platform that lets you run and manage AI agents directly on your own hardware. It keeps your data private while allowing you t

@afterquery/compute

TypeScript SDK for the AfterQuery compute API — isolated sandboxes for AI agent pipelines

@rulvar/executor

Rulvar isolated tool executors: reference ToolExecutorProvider adapters that run tool work out of process (subprocess and container) so hostile or model-generated scripts cannot reach host capabilities.

agent-orchestration-process

Run bounded, non-interactive jobs across agent harnesses with isolation, exact resume, and provenance

agentbox-runtime

Policy-sandboxed runtime for AI agent processes: enforce a policy file, capture a tamper-evident trace, deterministically replay any run.

brinqz-agent

Secure local execution agent for Brinqz Cloud

langchain-opensandbox

OpenSandbox sandbox backend for LangChain deep agents

sbclaude

Run Claude Code in a throwaway Docker container.

aily-lex

Headless AI Agent Runtime — multi-host, MCP-compatible

agentway

Python SDK for AgentWay - runtime control plane for AI agents

arkady-agent

Arkady's standalone agent runtime: hosts Agenton-composed Pydantic AI agent runs behind a FastAPI server, with knowledge, ask-human, and sandboxed shell layers.

pyagentos

The Operating System for AI Workers — build, deploy, and manage AI agents.

pydantic-ai-daytona

Daytona sandbox capability for Pydantic AI agents

@tribu-do/workbench-cli

AI-first sandboxed agentic development workbench — task orchestration, hierarchical memory, sandbox mediation, multi-agent parity, pluggable deployment.

@ludecker/agentic-bridge

Agentic OS local bridge — run dispatch, phase runner, Cursor adapter

openpond-sdk

Server-side TypeScript SDK for OpenPond sandboxes and agentic work

@aimform/compute

Aimform Compute — sandboxed tool execution, agentic actions, and workflow automation.

agent-cow-postgresql

PostgreSQL Copy-On-Write for AI agent workspace isolation

agent-substrate

Durable Agents for Agent Swarms, LLMs, and Autonomous Systems

k8s-agent-sandbox-awx

Broker-backed, kubeconfig-free overlay for the Agentic Sandbox client.

k8s-agent-sandbox-base

A client library to interact with the Agentic Sandbox on Kubernetes (seamed republish).

langchain-sandbox-opensandbox

OpenSandbox sandbox backend for LangChain Deep Agents — run agent-generated code and file operations in an isolated OpenSandbox environment.

tianshu-agent-os

Tianshu - Async AI Execution Platform

disk

Pure-JS client and CLIs for Archil disks and sandboxes

@ai-sdk/code-mode

QuickJS-backed code mode tool for AI SDK

@cofy-x/axern-sdk

TypeScript SDK for Axern agentic infrastructure and isolated sandboxes

ghemud-agentkit

A secure, provider-agnostic Python runtime for AI agent tool execution with policies, budgets, approvals, and audit trails.

ifed-agent-abi

IFED agent-side ABI: the ctypes bridge to the compiled C++ libtorch runtime

langchain-hyperlight

LangChain tool for executing untrusted code in Microsoft Hyperlight micro VMs (experimental, AI-generated)

langchain-microsandbox

Microsandbox integration for Deep Agents

openshell-agent-runner

Launch ephemeral agents for single tasks in OpenShell sandboxes.

sandbox-agents

Python client for the Sandbox Agents control-plane: agents, sandboxed sessions, streamed runs.

hostagentics

HostAgentics review: compare features, pricing, use cases, access model, and alternatives for this AI Agents Platform agent in 2026. Managed hosting for workflow engines and autonomous AI agents

agentboxer

Run coding agents in a per-project dev container sandbox.

rlm-agent-runtime

Kernel-side runtime shim for Zero recursion

sagent-code

AI agent that streams JavaScript into a persistent sandboxed Node.js runtime

qubicl-cli

Local, observable computers for any compatible AI agent

sandboxedjs

A Linux-like container that runs entirely inside Node.js — POSIX shell, ~140 coreutils, Node.js and Python runtimes, virtual filesystem and networking. No Docker, no VM, no native modules.

@zvada/agent-server

Harness-agnostic agent execution engine: run Claude Code, Codex (app-server), and any ACP agent behind one interface with a normalized event stream, multi-turn sessions, and resume. Root export is the wire contract; /core, /server, /client are the seats.

@tva-agentic-design/runtime

Local HTTP/WebSocket runtime and production Studio server for Agentic Design Runtime.

More in Agent frameworks & infrastructure