#cogo ai responses and user messages have markdown rendering now

eliminating that top-right "new pr session" button – soon, when any #cogo-managed claude code session opens a PR, it will create this pr badge you see on the bottom

pushing messages to interrupt the claude code session was still clunky. now it works smooth with instant responsiveness. #cogo the goal is to get claude code that runs on any node of yours –e.g. your local "quarantine" mac mini/studio– as nice to control from the cloud (cogo.dev) as in its direct terminal UI

first worker creation in onboarding flow of #cogo

feature dependencies in #cogo! (a feature will not get worked on until another is done)

interrupt claude's work with your latest message now just means pressing "enter" a second time in #cogo

when ai sessions create PRs they are now visible as seperate pills that you can mark as merged or delete in #cogo

#cogo the UI for high-parallel AI development / agent orchestration is nearing product ready, i.e. not just for me, but also for thee. if you have a mac mini computer in a corner somewhere that runs your agents this will be a cloud based orchestrator and UI for running those agents. you'll just control them from cogo.dev.

#cogo will not just help you execute your next step; it will fill the gap between your goal and your next step

started making #cogo self-explanatory for new onboarders, then Claude mysteriously early choked, "session limit hit". have not hit that for a while, have they changed their session limit? next feature of #cogo: auto-continue work when the session limit lifts

#cogo sessions can now create skills of their own accord mid-session

did some name storming for #cogo

#cogo is becoming a very well-oiled machine for developing with ai hyper fast

Added: basic PR integration into #cogo sessions

Thus also fixed: editing session display names in #cogo

little bugfix on name editing #cogo

My cloud based claude code orchestrator codenamed #cogo has reached a point where developing the thing itself using #cogo is the easiest way to go forward, so that’s what I’ve been doing today, along with a major update on #happenlist (also done in #cogo) where I can now steer event collection using prompts and mass-populate the index queue

whooo automatic worktree resolution is working in #cogo one thing less to do in the highly parallel software-pumping process

greatly improved the hands-on cloud UI for claude code sessions in my grand cloud based claude code orchestrator #cogo. Also writing messages that interrupt the claude code session works now.

can't stand building a serious ai setup with dozens of clis, authentications/tokens/secrets and instruction documents, all one a single machine. needs to be in cloud. otherwise i will have to do all of this again when i get a different computer! continuing work on my cloud claude code orchestrator #cogo