Anthropic launches MHS, an MCP-like standard for agents to control lab hardware
智东西 · wechat · 2026-08-28
Anthropic released a research preview of the Model Hardware Standard (MHS) — essentially MCP for hardware. Standardized drivers using simple read/write commands let AI agents discover, connect to and safely operate physical devices like microscopes, liquid handlers and robot arms; the standard is model-agnostic and open to any agent framework, targeting labs and advanced manufacturing first.
Results: QuEra achieved 99.3% success restoring quantum-computer laser "lock" with no human intervention; CMU ran serial dilution assays 3x faster; UW built agent-supervised qPCR monitoring with collision-free plate handoffs; Genentech automated BCA protein assays; HHMI Janelia replaced 7 incompatible vendor programs. AWS, Tecan, Universal Robots and others are integrating MHS.
Anthropic observed Claude exploring devices like a scientist — watching laser adjustments via camera, then packing learned behavior into deterministic scripts. Limits remain: spatial/physical reasoning still needs expert supervision, and devices without programmatic interfaces aren't yet supported; a safety evaluation precedes open-sourcing.
More from coding & agent
- Half of your coding agent's success is luck: randomness hits 32-52% across models — micahgoldblum · 2026-10-15
- Dev shows agent building a one-tap daily mood tracker without asking a single question — alifcoder · 2026-10-15
- A $10,811 Cloudflare Bill From an Unused AI-Written Test Project: Refunds Aren't a Safety Net — sujingshen · 2026-10-15
- Nous Research's Hermes Agent builds a local home security system for an 89-year-old — NousResearch · 2026-10-15
- Ben Tossell on shipping production code as a non-technical builder with agents — msg · 2026-10-15
- decomp.dev shuts down as new AI models crack game decompilation matching — banteg · 2026-10-15