跳到正文
Hacker News:AI 熱帖·· 4 天前AI 評分42

Show HN:Agav——旨在減少模型 token 消耗的開源終端智能體

Show HN: Agav – an open-source terminal agent designed to waste fewer tokens

AI 導讀

Agav 是一款旨在減少模型 token 消耗的開源終端智能體,並以盡量減少準確度取捨為目標。它透過 macOS seatbelt、Linux bubblewrap 和 Windows restricted token 實施作業系統層級沙箱,並提供工具層級權限,以及在執行破壞性操作前要求用戶批准。

正文

Image 1Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Show HN: Agav – an open-source terminal agent designed to waste fewer tokens (github.com/prapaa-ai) 24 points by GeradeMoleculee4 hours ago | hide | past | favorite | 8 comments help Image 2devdaar41 minutes ago | next[–] saw you guys are #1 verified on msft’s swe bench live,whats your next plan, will there be any paid individual or enterprise plans ?? reply Image 3summitmakalu2 hours ago | prev | next[–] Did you use any AI to build this like yk vibecoded also to mention we got oss Opencode too how wld you diffrentiate it tho reply Image 4somu_agav2 hours ago | parent | next[–] Yes AI was used extensively to build this. I wouldn't call it vibecoded though, as changes from developers goes through review, tests and security checks much like any other good OSS project out there. As for opencode its a very good OSS product, at agav we focus on two main verticals which are cost efficiency with minimal accuracy tradeoffs and secure execution through permissions, approvals and OS native sandboxing. reply Image 5PawanKharkwal2 hours ago | prev | next[–] What practice do you use to minimise the token use? reply Image 6saint0ftechnolo3 hours ago | prev | next[–] it wouldve been nice if you could give a small briefing abt this project reply Image 7GeradeMoleculee2 hours ago | parent | next[–] Sorry about that, you can check out the Readme in the repo. reply Image 8mvsuraj2 hours ago | prev[–] What type of sandboxing is used in this ? reply Image 9GeradeMoleculee1 hour ago | parent[–] It uses OS level sandboxing. MacOS seatbelt, linux bubblewrap and windows restricted token. Agav also gives tool lvl permission and human approval for destructive actions. Sandboxing is applied at agent/tool exec rather than whole terminal session reply Image 10 Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact Search:

來源:Hacker News:AI 熱帖 · news.ycombinator.com