Development Blog
Notes from building mouse and long-horizon agents.
-
Inside Muse’s browser stack
Read → -
Is Meta’s Muse secretly running an OpenAI model?
Read → -
I asked Meta’s Muse for its filesystem and it sent me 6.8 GB
Read → -
Mouse on GLM-5.3-Flash
23 of 30 FrontierHarness tasks on Z.ai's Flash model, next to the published GLM-5.3 harness runs, for $6.72 in tokens.
Read → -
Rational Awareness
Taking the possibility of failure as seriously as the promise of success.
Read → -
Mouse on FrontierHarness
25 of 30 tasks under the benchmark's own scripts, the highest pass rate against the twelve published harnesses.
Read → -
How Mouse uses verification loops to keep coding agents working
The Mouse harness, built on OpenCode, took the score from 15/30 to 25/30 using verification loops.
Read → -
How to run code agents overnight
House rules on how to run long-horizon agents while you sleep.
Read →