Running in production · Independent · 23 Oct 2026

e2b-dev/runtime

The runtime behind every E2B stack: Cloud, Enterprise, and your own machine.

← Running in production
E2B Runtime: AI sandbox engine. Runs isolated Firecracker microVMs for AI agents.

In simple English

This program runs the hidden engine behind secure AI coding environments. It creates tiny, isolated Linux computers that start in milliseconds from saved snapshots. Your AI agent can run untrusted code inside these boxes without breaking your main server. It pauses when the agent stops and wakes up instantly when new work arrives. Running it yourself requires a dedicated Linux server with virtualization support.

What it is

E2B Runtime is the open-source backend for cloud agent execution, combining a control plane API, node orchestrator, edge router, and virtual machine agent. It uses Firecracker (a lightweight virtualization technology used to run secure virtual machines fast) and Go to manage fast-resuming sandboxes.

What it is good for

Platform teams and solo developers who want to run AI agents inside secure, isolated sandboxes on their own infrastructure without relying on a hosted cloud service.

Stars1.8k
LicenceApache-2.0
Latest2026.30 (2026-09-10)checked 23 Oct 2026
Last code change2026-10-23
Written inGo
Setup effort
4 / 5
Payoff
4 / 5
Good
  • Boots sandboxes instantly by restoring memory and disk snapshots instead of booting a full kernel.
  • Provides strong hardware isolation using microVMs, cgroups, and network namespaces.
  • Includes built-in support for pausing idle sandboxes and waking them transparently on incoming traffic.
Watch for
  • Requires a dedicated Linux host with KVM (Kernel-based Virtual Machine, a technology that lets your CPU run virtual machines directly) enabled in the hardware.
  • Complex multi-service architecture needs databases like PostgreSQL, Redis, and ClickHouse to run fully.
  • Setup and local development require familiarity with container composition and Go tooling.
Replaces

Manual virtual machine provisioning and slower container-based sandboxes for AI agents

Works well with

Docker Compose

Similar repositories
  • modal-labs/modal-client527 stars

    A platform for running serverless functions and containerized workloads in the cloud with fast startup times.

Also reviewed on Repo Radar
  • vectordotdev/vector23k stars

    One collector that gathers logs and numbers from machines and sends them to one or more monitoring services.

  • bluewave-labs/Checkmate11k stars

    A self-hosted monitor that checks whether sites and services answer, records response times and sends alerts.

  • statping-ng/statping-ng2.0k stars

    A single-file status page and uptime checker that works with SQLite, PostgreSQL or MySQL.

Install
mkdir e2b && cd e2b
curl -fsSL --remote-name-all "https://raw.githubusercontent.com/e2b-dev/runtime/main/embed/compose/{compose.yaml,.env}"
docker compose up -d --wait

Independent: reviewed for Repo Radar outside the daily editions. Every independent repository

On Stack Radar: Build & Ship › Developer Environments › Cloud Development.

Get the next edition in your inbox

A dozen repositories, opened and checked. The licence read, the last release dated, and the ones that did not make it named with the reason. It is the half most lists leave out.

No tracking pixels. One click to leave. The archive stays free either way.

We use your address to send the edition and nothing else. Confirm by email, leave in one click. How we handle it.