Skip to content
All features
AI runtimepackages/ai/sandbox-runtime

sandbox-runtime package

Vercel Sandbox + Lambda + local runner adapter under one execute() call. Picks the runner from a policy rule list.

Open docs
Stability
Stable
Scope
Global
Boundary
packages/ai/sandbox-runtime

sandbox-runtime

packages/ai/sandbox-runtime

AI runtime · tool registry · MCP boundary

Ready

Capability graph

ai/sandbox-runtime
sandbox-runtime
Tool policy
Stream UI
Output

Run ledger

  1. 1
    Input
    packages/ai/sandbox-runtime/request
  2. 2
    Plan
    message reducer -> model turn
  3. 3
    Tool policy
    tool schema · auth scope · rate limit
  4. 4
    Stream UI
    events -> UI state machine
  5. 5
    Output
    @nebutra/agents response envelope
p50 latency
190 ms
events/sec
529/s
providers
7
eval score
95
Usagesandbox.ts
typescript
sandbox.ts
1import { createSandbox } from "@nebutra/sandbox-runtime";
2
3const sandbox = await createSandbox({
4  runtime: "node22",
5  timeoutMs: 30_000,
6  network: "egress-only",
7});
8
9const result = await sandbox.run(code, {
10  inputs: { user, plan },
11});
More in domain · AI runtime
3d-pipelinePackage
packages/ai/3d-pipeline

3D asset generation + render — text-to-mesh + USD export + Three.js preview, with GPU-pool routing on the backend.

Explore
agent-runtimePackage
packages/ai/agent-runtime

Lower-level agent execution — message reducer, tool invocation, streaming state machine. The kernel @nebutra/agents wraps.

Explore
agentsPackage
packages/ai/agents

Multi-step agent runtime — Vercel AI SDK foundation, tool registry, streaming UI, MCP integration. The skeleton for every Sailor agentic feature.

Explore
ai-primitivesPackage
packages/ai/ai-primitives

Shared helpers used across all AI packages — token counting, prompt templating, stream-parser, retry-with-backoff. No provider dependency.

Explore