A TypeScript runtime package for executing AI agents inside sandboxes. Version 0.15.1 (active development) provides a WebSocket-based server that handles agent lifecycle: receives invocations, runs handler, manages artifact uploads, and returns structured results. Built with strong TypeScript types. Key differentiators: designed for sandbox environments, supports OpenAI and Claude built-in handlers, integrates with a Python client for external orchestration. Ships types, ESM and CJS compatible.
npm install runtimeuseNo compatibility data collected yet for this library.
Verified import paths — ran on the pinned version, not inferred.
Shows minimal setup: import server and handler, create instance, start listening.
Replace import RuntimeUseServer from 'runtimeuse' with import { RuntimeUseServer } from 'runtimeuse'.Ensure claude CLI is installed in sandbox if using Claude handler.
Set OPENAI_API_KEY in sandbox environment before starting runtime.
Use import { RuntimeUseServer } from 'runtimeuse' instead of import RuntimeUseServer from 'runtimeuse'.Provide a handler: import { openaiHandler } from 'runtimeuse' and pass { handler: openaiHandler }.Install @openai/agents package or use a different handler.