Resat Ugur Ulu

Building @graphyn_

[Resat Ugur Ulu @resatu]

Stay small enough long enough, you’ll be big enough soon enough.

[Resat Ugur Ulu @resatu]

@Aditiiitm System components are knowledge, agents, modes, and machines. Relaunch is essentially rehydration, with desktop agents running as separate PIDs via harnesses.

Basically, the system turns your machine into a federated node, allowing you to port any agent with any configuration.

[Resat Ugur Ulu @resatu]

@Aditiiitm That's one of the leanest parts of the system. Threads are moved mainly by forking sessions, copying the JSONL, and relaunching. Anyone can do it, but it's not the usual way of working, and devs are lazy, so it becomes a problem that can be solved with proper runtime design.

[Resat Ugur Ulu @resatu]

Codex limit reset feels like

[Resat Ugur Ulu @resatu]

@w1nt3r_eth that affordance needs work

[Resat Ugur Ulu @resatu]

just teasing https://t.co/MhicTDsX2r

[Resat Ugur Ulu @resatu]

@realameerdev capture-to-agent runtime: https://t.co/FLKJLkyRcK

[Resat Ugur Ulu @resatu]

@fkadev @MustaAras @steipete Real claws never share their secrets...

[Resat Ugur Ulu @resatu]

@MustaAras @fkadev @steipete POV: you've instantly got clawpilled

[Resat Ugur Ulu @resatu]

@mega__d I'm looking for teammates who are excited about portable AI agent runtimes and distributed systems. Ideally they have experience with agents, dev tools, or building desktop apps.

[Resat Ugur Ulu @resatu]

@KscYusuf @blyth_ai of course

[Resat Ugur Ulu @resatu]

@finchbuilds @Amaar_Ali12 would value that

[Resat Ugur Ulu @resatu]

@dhavaljoshi07 capturing agent work: https://t.co/FLKJLkyRcK

[Resat Ugur Ulu @resatu]

@finchbuilds @Amaar_Ali12 early private builds soon

[Resat Ugur Ulu @resatu]

@YashHustle_22 prompt-to-agent runtime: https://t.co/FLKJLkyRcK

[Resat Ugur Ulu @resatu]

@shahrukhghazaan @claudeai both, with manual review

[Resat Ugur Ulu @resatu]

@husimorning @X useful signal, thanks

[Resat Ugur Ulu @resatu]

@finchbuilds @Amaar_Ali12 task rubrics matter