Scale computer-use 2.0 with open-source drivers, cross-OS sandboxes, and benchmarks for training, evaluation, and data generation.
Rust agentai-agentapplecomputer-usecomputer-use-agentcontainerizationcuadesktop-automationhacktoberfestlumemacosmanusoperatorswiftvirtualizationvirtualization-frameworkwindowswindows-sandbox
docs(rfc): record the released-Driver GTK3 save-note result for RFC 4268 (#4315)
Francesco Bonacci4243157a7bd2
diff --git a/rfcs/4268-jev-use-native-candidates.md b/rfcs/4268-jev-use-native-candidates.mdindex 5c3bf8807e..9cf1b433d7 100644--- a/rfcs/4268-jev-use-native-candidates.md+++ b/rfcs/4268-jev-use-native-candidates.md@@ -851,8 +851,9 @@ Phase 3 live evidence (released Driver 0.30.2, cua-perception 0.2.1,`cua-s1-4b-0.2@16818868`; each run verified by the harness's own state file):- Live Jev passed 5/5 in every native row (AppKit, WPF, GTK3 × three tasks- × Python and TypeScript) except GTK3 `save-note`, blocked by #4291 until- Driver 0.30.3 ships.+ × Python and TypeScript). GTK3 `save-note` ran on released Driver 0.30.4,+ which includes the #4291 fix, from the canonical Linux installer: live Jev+ and S1 each passed 5/5 in Python and 5/5 in TypeScript (20/20 runs).- S1 missed six of 82 native runs: three `counter` runs where it abstainedafter two increments (the value-free request makes it count from history),two macOS `save-note` runs where it pressed Save before typing, and one@@ -867,5 +868,5 @@ Phase 3 live evidence (released Driver 0.30.2, cua-perception 0.2.1,and S1 in 211/216.Acceptance criteria not yet met, which keep this RFC `accepted`: S1 5/5 in-every native row, GTK3 `save-note` on a released Driver, WinUI3 coverage,-and accuracy at about 12 and 24 candidates.+every native row, WinUI3 coverage, and accuracy at about 12 and 24+candidates.