Holy crap, who the hell did this?!!
@ArchiveExplorer13K followers
17,492 views
17 Comments
Comments from the original post on X (Twitter)

@itsthedonhashim
@ArchiveExplorer careful with dependency hell. Sometimes these setups need way more config than advertised.
1 like

@crypto_ninjas
Impressive plug‑and‑play AI brain, love it
1 like

@ArchiveExplorer
it genuinely makes me happy that this is useful to you

@kraayenJon
That’s cool! You should post it in
1 like

@Slonski_rt
the expensive model finally gets to do the thinking
1 like

@Shtihmas11
Claude Code and Codex support
1 like

@RuujSs
There are always bunch of good devs catching the trend and making it better
1 like

@rewind02
saving Hermes-Jev-Skills mate
1 like

@ArchiveExplorer
i'm really glad this is useful to you what are you building?

@gusik4ever
almost 1k stars on repo sounds promising bro
1 like

@ArchiveExplorer
once you start using it give me some feedback curious how you find it

@catmanyau
This is a dispatcher for agents: cheap routing decisions keep the expensive model focused on the actual work, while safe-action boundaries limit what can happen automatically.

@ggwplabs
nice that the local skill layer avoids any token spend on the model itself and you can push decision cost under three hundredths of a cent per call

@teddessert
Not interesting until someone compares it with another memory tool using benchmarks.

@sophsbeats911
Interesting perspective

@sin_management
Will give it a try ..my BUZZ multi agent setup is getting out of hand.

@hx0000001
如果 Jev 本身是付费云 API,那 'API key 不接触 agent' 更像是把决策路由转移到云端那一层,模型选择和记忆裁剪都在云上发生,延迟和成本在长任务里怎么权衡?