ldraw-nova: AI agents build LEGO models by writing Python geometry
In ldraw-nova, models write a plan and a Python generator instead of raw coordinates, but nothing checks physics or stability.
DigitalOcean Managed Agents: stop paying for idle coding agent time
DigitalOcean Managed Agents pause microVM billing while sessions sit idle; check idle time, secrets, resume behavior and quotas before moving agents.
Huseyin Kaplan · 1 min read
A coding agent spends much of its session waiting. On a test run, a tool call, a human reply. The machine under it bills you for every idle minute anyway.
DigitalOcean opened Managed Agents to everyone as a public preview this week. The news isn't a new model. It's where the agent runs.
What's in it:
- Harness Runtime: a microVM with a browser and a coding sandbox. Sessions pause, resume and fork with files and context intact. CPU and memory charges stop while paused.
- Action Gateway: one MCP endpoint for thousands of tools. Credentials are brokered at execution time and never reach the model.
- Your own harness: Claude Code, Codex CLI, OpenCode, LangGraph.
Before you move an agent there, check:
- How long your sessions actually sit idle.
- Whether secrets live in your prompts today.
- What a resumed session really remembers.
- Preview quotas and rate limits.
Agents work in bursts. Pay for the bursts, not the waiting.
