Meta's Muse checks a training run and locks the Mac remotely — agent-computer control is mostly permissions work
OldChemical3853 · reddit · 2026-09-25
A developer built a small bridge so Meta's Muse agent can act on his Mac from his phone, and recorded a real session: Muse reads the Terminal, reports the epoch, loss and accuracy of a running training job, then locks the machine on request.
His takeaway: the hard part of letting agents reach your actual computer is permissions, not intelligence — per-action consent, small window captures instead of a live feed, and rejecting stale observations before any input.
The post doubles as promotion for a free beta, but the demo shows what agent-on-real-computer control looks like in practice and where the engineering effort goes.
More from coding & agent
- Perplexity CEO shows agent building browser-playable 3D zombie shooter in hours with zero 3D assets — AravSrinivas · 2026-11-11
- The complete harness engineering roadmap: how OpenAI shipped ~1M lines with zero hand-written code — blaizedsouza · 2026-11-11
- Micius-Flash: 27B Multimodal Model for Hardware and Semiconductor Agents — zhiman-ai · 2026-11-11
- Viral 'the harness is the machine' thread touts unverified Claude Fable/Mythos 5.2 9-equation architecture — blaizedsouza · 2026-11-11
- Compaction creates a new failure mode for coding agents: critical bug details can vanish mid-fix — blaizedsouza · 2026-11-11
- What 40+ coding-agent harnesses do between your prompt and the model, dissected — blaizedsouza · 2026-11-11