A one-shot coding agent: send it a prompt, and Claude Managed Agents runs the
agent loop while a worker in an ephemeral Render Sandbox runs its commands and
file edits. Files the agent writes to /workspace/output are copied back to
your machine.
Render Sandboxes is in early access. Claude Managed Agents is in beta and uses the
managed-agents-2026-04-01API.
Follow the tutorial: Run Claude Managed Agents on Render Sandboxes.
You need Node.js 22+, a Render workspace with Sandboxes enabled, Claude Managed
Agents access, an existing Managed Agent, and a dedicated self_hosted
Environment with a Console-generated Environment key.
npm ci
cp .env.example .env # fill in your credentials
npm run agent -- "Create a hello world HTTP server in /workspace/output. Test it before you finish."Options:
--timeout <seconds> Override the 600-second deadline
--output <path> Copy /workspace/output files to this local directory
Each run writes run.json, events.jsonl, and result.txt under
runs/<run_id>/. The runner deletes only its per-run Session and sandbox. It
never deletes the shared Agent or Environment.
@anthropic-ai/sdk0.135.0@renderinc/sdk1.3.0ant2.1.0 for Linux amd64 and arm64, checked against a pinned SHA-256
npm run check # typecheck and tests