A weekly publication of the top three papers from arXiv last week. Papers with arXiv v1 published 2026-09-21 through 2026-09-28 UTC.
Open by default_
A security research team building open review workflows.
Midkernel started as a tool for our own security research: running open security workflows against real codebases without rebuilding the plumbing every time. The plumbing — sandboxes, model access, artifacts, and a way to tell which workflow was worth running — turned out to be the product.
The workflows were never the secret. They live in a public repository, written to be read before they are run. What was missing was a place that ran them for you, kept the whole trace, and showed the price before you started.
So that is what we build: open workflows, a hosted run you can read end to end, and — as it matures — a public evaluation of how well each workflow performs. Everything that can be open source is. The company earns the difference between what a run costs and what it is worth to have it be one button.
Show the log.
A finding without its trace is an opinion.
Public over private.
Workflows, benchmarks, the runner, the methodology. Closed only where a customer's code requires it.
Price first.
You see what a run costs before it starts, and the invoice matches.
Small on purpose.
A research team, not a sales floor. Growth comes from being the obvious way to run open work.
A weekly publication of the top three papers from arXiv last week. Papers with arXiv v1 published 2026-09-14 through 2026-09-21 UTC.
A weekly publication of the top three papers from arXiv last week. Papers with arXiv v1 published 2026-09-07 through 2026-09-13 UTC.