yesh

bish bosh kiri, the niri style agent orchestrator ive been making is shaping up real nice - supports codex, claude, pi, opencode - keyboard first - mcp support so agents can control it - scratchpad - drop ideas in and start fresh sessions - in chat edit diffs, session diffs using @pierrecomputer

you know, I feel this. I realized the last few days that it is so hard to communicate and I think at least it’s a trap of not failing if no one sees it. What you said about being brave is spot on. I have days where i plow ahead and share and then feel exposed There is a lot of my pov here. But, that feeling has been something I’ve gotten very familiar with

guys, i honestly don't know wtf is happening to me i started sharing my work with you again and you like it and respond to it, great now how tf do i make myself continue everything screams STFU!!! do you know the amount of shit i've built since last year and havent shared ONE SCREENSHOT for whatever reason???? i shared some of them during shipmas and i stopped. i have probably like 20 more products ready. idk how this sounds but i have some versions of products like trustmrr, openclaw, clicky, agentic browsers, agent orchestrators blah blah blah with commit date proofs of when i started building them.. seeing someone else announce something similar to what you've been building SiLeNtLy is INFURIATING ngl. because they were brave. it's all it takes. i just moved on from every single product as soon as it felt like it was ready to be wrapped up and shared fuck me i've built an ai browser for productivity in january on 2025 and i haven't even shared the name of it idk how i keep falling into this hole of not communicating... i think it's easier to be like "haha my products fail because i'm too scattered and working on too many things", it's like a safety net also i HATE that i'm a visual person but NOT recording any video/audio most of the time.. it just feels like a waste of time compared to sHiPpInG, ugh, i get so nervous when i'm not clicking on a keyboard just coding infinitely, without sharing anything fucking idiot

got nerd sniped by @steipete into building my own cotypist - uses pi as base - it supports both local modals and everything else pi supports - right now, i have it only trigger on a keybind. im not too thrilled about having it run constantly I have a couple of ideas of improving the base. But, a few hours in its pretty alright

You don’t have to live this way Caffeinate on Mac or better yet An old laptop or vps running quietly somewhere

@lexrus They done have to x.com/yesh/status/20…

All these mac apps look like the same. This is the new GitHub Copilot app.

I am terrible at showing what I made but, kiri now has terminal support (it has for a bit there) so that I can use claude code without interruptions It still remains keyboard first even with this and also supports chained workflows and orchestration with kiri mcp/cli

I’ve been doing something similar for long running goals in both codex and Claude I add a LOG.md and then once it’s all done make a detailed interactive html letting me explore the changes

kiri - my keyboard first agent orchestrator has a website now in my recent iteration i've added scratchpad integration and orchestration using mcp mostly inspired by solo kiri-8zk.pages.dev

@floguo Just make your own, surely the world can do with one more x.com/yesh/status/20…

okay so TOOLS: superset, conductor, wonder, rivet, paper, endless tools, unicorn studio, noon, magic path, magic patterns, v0, lovable, github copilot (wake up babe new desktop app just dropped!), cursor, antigravity, windsurf, devin... what are you reaching for these days?

Nothing like spring in Toronto

In codex, if you are using goal. I found the better way to make it stay on track and do what I want is to keep having a discussion with it. Set constrains and verification loops Then let it leverage `create_goal` in app server and let it create its own goal

I think a lot of us were trying to get a handle on this review-feedback workflow around the same time a few months back. But, the other ones I really liked are - @iamsahaj_xyz lumen - @bentlegen hunk - @plannotator diffs

I have been using quickdiff for months now - vim motion - annotate and comment and get ai to fix them - --pr flag to see any PR and review it. approve/decline flow - git/jj support - theme support github.com/Yeshwanthyk/qu…

@mweinbach Not alone at all. It was just being dumb all day long.

I’m not trying to be that guy But GPT 5.5 feels significantly worse today. I don’t think I’m alone with that one.

Make your own wrapper. Pick you poison for the complexity you need use a vps + tmux EC2 + an agent worker webui + opencode + daytona/modal cloudflare workers + agents sdk + durable objects

alright kiri now has "terminal" mode and supports all terminals including claude code there is some jank with moving between sessions. just need to iron it out

@thdxr Debase with a rebase

im not an honest person i squash my commits

Ok it was not just me thing. I thought I was going insane.

Before using the next version of Claude code