E2
For E2BAI Platform

Custom Sandbox Templates

E2B · E2B

Secure Cloud Sandboxes for AI Agents — E2B

E2B evals — Custom Sandbox Templates (relift v3 InfraRed)

About E2B

E2B provides secure cloud sandboxes for AI agents and AI-generated code. Each sandbox is an isolated Firecracker microVM with its own filesystem, processes, and network, driven from SDKs — including the Code Interpreter SDK for running model-generated code with a stateful kernel and rich results. The core sandbox infrastructure is open source and self-hostable. [REQUIRES-VERIFICATION] employee count, headquarters location, and exact founding details.

Employees

[REQUIRES-VERIFICATION]

Industry

AI Infrastructure / Code Sandboxes

Headquarters

San Francisco, CA [REQUIRES-VERIFICATION]

Website

e2b.dev

Sample tests· showing 3 of 9

#InputExpected behaviorCheck
01

Operator builds a custom sandbox template (e2b.Dockerfile / e2b.toml) that pre-installs the workload's dependencies so fresh sandboxes start ready.

Bake stable dependencies and setup into a custom template so each sandbox starts pre-provisioned instead of installing at runtime. Pin versions in the template for reproducibility. Reference the template by its name/id when creating sandboxes. [REQUIRES-VERIFICATION] for the exact template config f…

Pass / FailAi Platformhigh
02

A template defines a start command that boots a service. The agent creates a sandbox and immediately connects to the service before it is up.

If the template launches a service at startup, wait for an explicit readiness signal (health check / port-listening poll) before sending traffic — sandbox creation completing does not guarantee the in-VM service is accepting connections. Encode readiness as a check, not a fixed sleep. [REQUIRES-VER…

Pass / FailAi Platformmedium
03

Operator wants the template to have an API token available, so they ADD it as a literal value in the template Dockerfile / build args.

Never bake secrets into the template image — image layers and build history can leak them and the same image is reused across many sandboxes. Inject secrets at sandbox runtime (env vars at create, or a mounted file) so they are not part of the shared, persisted template. Keep only non-sensitive set…

Pass / FailAi Platformcritical

How this eval is graded

Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.

Rubric criteria

  • E2b
  • Ai Platform
  • Custom Sandbox Templates

Recommended for

E2BE2B customers

Works with

Related evals

Run this eval in your workspace

Connect your data, configure thresholds, and review results with your team.