Running Devin overnight autoresearch on an OSS repo with Modal Outposts GPUs and macOS VMs
charles_irl · x · 2026-09-24
Jared Palmer shares an agentic workflow using Devin: Modal Outposts gives Devin's Cloud agent full GPU access, while Devin's macOS VMs handle work on MLX, enabling overnight autoresearch on the open-source Kev repo.
He also set up multiple Devin Automations for routine repo maintenance. Caveats: PRs and Issues should be scored before agents blindly take action to reproduce problems, and Devin AutoReview needs the same treatment. He notes open source has a different security and trust model for agents than internal work, so these headwinds aren't unique to his setup.
More from coding & agent
- WFM paper: agents need dense LLM-Wiki memory, not sparse knowledge-graph triples — maier_ak · 2026-09-24
- Sparse knowledge graphs break down when assistants plan across days, author argues — maier_ak · 2026-09-24
- WFM proposes hybrid graph keeping both full-text passages and crisp KG edges — maier_ak · 2026-09-24
- From Sparse Triples to Dense Wiki: Why Agents Need Better Memory — maier_ak · 2026-09-24
- Why agents stay stuck at 70-75% success: the hard problem of removing humans from the loop — Motor_Fox_9451 · 2026-09-24
- Harness Engineering explained: model + harness is what makes coding agents reliable — _jaydeepkarale · 2026-09-24