Kelven Severo

Just building.

[Kelven Severo @KelvenSeveroF]

@lateascent i think plain users notice fast. the cursor jumping while you type is hard to miss, and that's when most people would just turn it off

[Kelven Severo @KelvenSeveroF]

Task success is the wrong score for computer use agents

Cua's bench only counts a task done in the background while you keep working. Best on macOS fully passed 58%. Codex computer use 33%.

If it needs your mouse, you didn't hand off the work. You just stopped yours.

[Cua @trycua]

1/ Today we're launching Cua Driver Bench, together with Scale AI's AgentEnv.

Can an agent do a desktop task in the background while you keep working undisturbed? On macOS, Cua Driver fully passed 58%, Codex computer use 33%, Vercel's agent-desktop 20%, browser-use 3%.

[Kelven Severo @KelvenSeveroF]

@ohheyyang yeah the split matters more than the score. a stuck agent costs you a minute, a confident wrong one costs you an afternoon

[Kelven Severo @KelvenSeveroF]

@markfenner @qoder_ai_ide the leftover worktrees and docker stacks bit is underrated. multiple agents leave a mess after the pr lands

[Kelven Severo @KelvenSeveroF]

@buildAgoat a date picker inside the thread beats describing a calendar to a text box every time. the card is the ui

[Kelven Severo @KelvenSeveroF]

@brendenvonhouse 180 links in llms.txt aimed at the agent is the right instinct. most docs still assume a human will click around

[Kelven Severo @KelvenSeveroF]

@VibeCoderOfek schema edits nuking the cache for the rest of the session is a silent bill. fresh thread after big schema moves

[Kelven Severo @KelvenSeveroF]

@iam_ihoudjedje mit keeps the binary. it doesn't keep your setup when the cloud kills the worktrees and secrets. export path first

[Kelven Severo @KelvenSeveroF]

@surlymochan claimed done without a replayable proof is just a green checkbox. independent evidence should be the default gate

[Kelven Severo @KelvenSeveroF]

@ishaansehgal @omnaraai recovery mid-command is the real gap. wait vs retry vs collect is where hosted harnesses quietly lie to you

[Kelven Severo @KelvenSeveroF]

@witcheer 142 combined is cool. for coding the number i care about is what one session gets when three others are also generating

[Kelven Severo @KelvenSeveroF]

@HamelHusain models sitting together at 74% on DeepSWE then splitting by 23 points after a filename tweak. the leaderboard became the product

[Kelven Severo @KelvenSeveroF]

@ganesh2412 yeah. if the revert path isn't what turns it red, the whole green-red-green loop is theater

[Kelven Severo @KelvenSeveroF]

@timhuyn in the case file itself. a sidecar the runner can forget to check is just a soft fence

[Kelven Severo @KelvenSeveroF]

@mksglu keeping raw tool output out of the context is the right call. does the agent know what's in the index or does it have to guess what to ask for?

[Kelven Severo @KelvenSeveroF]

@ctatedev a doctor command that measures the machine before picking settings is nice. does it learn different settings for CI vs local?

[Kelven Severo @KelvenSeveroF]

@brendenvonhouse a when to use clerk section in llms.txt is basically SEO for agents. pricing.md is the clever part, agents can't pick what they can't price

[Kelven Severo @KelvenSeveroF]

@_sumeetc agents saying done with no proof is the real gap. does it flag when the agent says tests passed but never actually ran them?

[Kelven Severo @KelvenSeveroF]

@BaselIsmail ran 9 of 10 times, robot walked 0 times. exit code isn't the test, the outcome is

[Kelven Severo @KelvenSeveroF]

@stretchcloud a 30 second vote works when 3 people are watching. at 2am with nobody around it's just a timeout. what does it default to, deny?