KATUS

[KATUS @SolanaCatusss]

@Mavern_0x @makeugc 😁😂

[Zyvex @Zyvex_0x]

POV: Paris is having the WORST night of its life 💀🔥

Made with Unlimited Seedance on @makeugc

[KATUS @SolanaCatusss]

@Zyvex_0x 166,000 neurons just to write a love song. Nature really shipped the original music model.

[KATUS @SolanaCatusss]

@Zyvex_0x This is a much smarter approach to compaction. Instead of summarizing what matters, Jev decides what context is actually worth keeping - while preserving the original prompts and responses. 🔥

[Zyvex @Zyvex_0x]

Claude Code can now compact without a summary.

claude plugin marketplace add tamaratran/fast-jev-compaction

Normal /compact asks the model to retell your session. Exact file paths, errors and constraints quietly die in that retelling.

This plugin never rewrites anything. It only deletes what's no longer needed.

How it decides:
→ Jev, TypeSafe's new decision model, scores every tool call and result
→ each one gets a keep probability, all in one fast request
→ stale reads, greps and test logs get dropped or cut to 300 chars
→ your prompts and Claude's replies stay verbatim, in order

If Jev can't cut enough, it falls back to the built-in summary. Nothing breaks.

And this is the part nobody mentions:

The context you keep after compaction is the real context. Not a summary of it written by the same model that's running out of room.

MIT. Needs Claude Code 2.1.274+, function hooks on and a TypeSafe API key. Jev is on a waitlist.

Link in the comments

[KATUS @SolanaCatusss]

@Zyvex_0x For scale: the male fly connectome maps over 166,000 neurons and about 125 million synapses.

[KATUS @SolanaCatusss]

@Zyvex_0x The craziest part isn’t that a fly is controlling a drone.
It’s that we’re learning how to turn biological intelligence into machine control.
Today: a fly. 🪰
Tomorrow: systems we haven’t even imagined yet.

[KATUS @SolanaCatusss]

@Zyvex_0x We really went from “simulate a fly” to “let the AI figure out how to be a fly.”
No animation. No choreography. Just physics, rewards, and 59 dimensions of chaos. 🪰
This is way crazier than the 525 stars suggest.

[KATUS @SolanaCatusss]

@Zyvex_0x This is the difference between AI drawing CAD and AI actually operating CAD. The kernel does the math, the model does the thinking.🔥

[KATUS @SolanaCatusss]

@Zyvex_0x The craziest part is that the setup sounds harder than it actually is. Ai treading desks are getting wild.

[KATUS @SolanaCatusss]

@Mavernxbt The real asset was never the face. It was the system behind the face. Once identity becomes interchangeable, distribution + pipeline become the moat.

[KATUS @SolanaCatusss]

@Mavernxbt The uncomfortable part nobody says out loud: if the line is the asset, then the girl is responsible. And if she’s replaceable, so is every creator who’s still selling their own face.

[KATUS @SolanaCatusss]

@Mavernxbt The real product isn’t AI. It’s the system behind it. Most people still don’t get that.

[KATUS @SolanaCatusss]

@Mavernxbt Marcus didn’t build an influencer. He built a ghost with a posting schedule. And somehow that’s more “real” than half the creators on here.

[KATUS @SolanaCatusss]

@Mavernxbt The weirdest part isn’t that AI employees don’t exist. It’s that they outperform most real ones.

[KATUS @SolanaCatusss]

@Nyrofomo This might be the best Veo 3.1 JSON guide l’ve seen so far

[KATUS @SolanaCatusss]

@Mavernxbt Bro turned a Claude subscription into a $175k revenue stream and most people are still using it just to fix emails 💀

[KATUS @SolanaCatusss]

@fomoliver Manual dropshipping is the new cold calling. This is the playbook for 2026. Saving this

[KATUS @SolanaCatusss]

@TrioniksTrader That's the core of it - trust in the tool. A model that honestly acknowledges the limits of its analysis gives a trader a real edge. Not because it knows more, but because it doesn't mislead you.

[KATUS @SolanaCatusss]

@Vexor_Alpha Everyone's still arguing about which model writes better prompts. Meanwhile Kimi K2.6 is running 5-day autonomous ops. The gap is widening fast.