AI runtimepackages/ai/code-execution

code-execution package

Sandboxed code execution for agents — JS / Python / shell — short-lived containers, network-isolated, capability-gated by @nebutra/permissions.

Open docs

Stability: Stable
Scope: Global
Boundary: packages/ai/code-execution
Owners: @nebutra/ai

code-execution

packages/ai/code-execution

AI runtime · tool registry · MCP boundary

Ready

Capability graph

ai/code-execution

code-execution

Tool policy

Stream UI

Output

Run ledger

1
Input
packages/ai/code-execution/request
2
Plan
message reducer -> model turn
3
Tool policy
tool schema · auth scope · rate limit
4
Stream UI
events -> UI state machine
5
Output
@nebutra/agents response envelope

p50 latency

128 ms

events/sec

637/s

providers

7

eval score

85

Usagecode-execution.ts

typescript

code-execution.ts

1import { codeExecution } from "@nebutra/code-execution";
2
3const result = await codeExecution.run({
4  tenantId: org.id,
5  // code-execution is part of the AI runtime — composable with other AI primitives.
6  input: payload,
7});

code-execution packagepackage

code-execution

Capability graph

Run ledger

code-execution package