Andrew Kalita

Took venture funding Spent 2 years building Company failed Made zero revenue Struggled with self-doubt for 2 years Started from scratch and small Looked for new opportunities Built a Chrome extension Focused on SEO Spent $0 on ads Made my first dollar Confidence returned Hit $25k in a year I am so grateful that slowly, but surely my builders energy comes back.

Internet was built for humans, agents just need a read-only database access.

I know that X has a lot of talented people, can someone do this please?

SEO, ASO, and app intelligence for the agent era - just curl, markdown, and 402.

@onusoz I that right to say that the need for this workflows is an artefact of current models not being able to follow instructions properly step-by-step?

acpx v0.4 ships Agentic Workflows, or as I like to call them "Agentic Graphs" It let's you create node-based workflows on top of ACP (Agent Client Protocol), to drive any coding agent (Codex, Claude Code, pi) through deterministic steps This let's you automate routine, mechanical legwork like triaging incoming PRs, bugs in error reporting, and so on... For example, OpenClaw receives 300~500 new PRs per day. A lot of them are low quality, but they still relate to real issues, so you have to address them somehow You need to: - extract the intent - cluster them based on intent - figure out if the proposed changes are legit, or whether they are slop local solutions, like trying to catch flies instead of drying out the swamp - if the PR is too low quality or the intent is not clear, close them - run AI review on them them and address any issues that come up - refactor them if the changes are half-baked - resolve conflicts - and so on... So that when the PR is presented to the attention of the maintainer, all the routine legwork is done and the only remaining thing is the decision to (a) merge, (b) give feedback to the PR author, or (c) take over the PR work yourself I wanted to build this feature since a couple months now, since Codex got so good. OpenAI models are now good at judging implementation quality, so I found myself repeating the same steps I wrote above over and over I also tried putting all this in a single prompt. But I believe there are workflows that should not be a single prompt, but a sequence of prompts in the same session That is because like humans, LLMs are prone to PRIMING. I claim that putting all steps in the same prompt at the beginning of the context will generally give suboptimal results, compared to revealing the intention to the model step by step Creating such a workflow also gives more OBSERVABILITY into the each step that an agent is supposed to take. Agent generates JSON at the end of each step, and that structured data can be used to monitor thousands of agents running at the same time in an easier way, on a dashboard Similar features have been introduced in e.g. n8n, langflow. But AFAIK they are not integrating ACP like the way I do I wanted to have a fresh approach, and to build an API that I can develop freely the way I want, so I created a new workflow API inside acpx The video is from the workflow run viewer, but that is not where you build the workflow. You build it by using the acpx flow typescript API. See examples/pr-triage in acpx repo Before building that, I started from a Markdown file with a Mermaid chart of the flow I had in mind. The Markdown file acts as a spec for the flow, and I have built the workflow through trial and error. I call this process "workflow tuning" I started working on acpx repo PRs one by one, tuning the flow, slowly scaling to more PRs. Finally, when I felt confident, I ran it in parallel over all external open PRs in the acpx repo. I believe it already saved me hours this week My next goal, if well received, is to set this up on a cloud agent so that it can process the 300~500 PRs the OpenClaw repo receives every day, in real time, as they come in I believe this will save all open source maintainers around the world countless hours and make it much easier to herd and absorb external contributions from everyone!

@forgebitz 23

last year we had some massive billboards next to the highway/airport if you guess how many signups we got from it you get a free hat

@snwy_me been there, now have this

i have no words

@gradientpull @clankoids Actually no, I was just testing it, not sure which one I opened.

@_avemeva_ @clankoids woah. keep in mind its not perfect yet, were at about 70% accuracy with the live pipeline right now, which is not that trustworthy. could you send me this post so i can check?

Installed @clankoids to check X bots replies. Entered the comments on random post. This is just freaking sad to see.

It feels like they have tested it on users like me first xD

@lydiahallie Is that possible to support mandatory use of skills by regex pattern?

Claude Code now supports an `if` field in hooks It uses permission rule syntax to filter when a hook runs, which is useful when you want a hook on some bash commands but not every single one!

What happened to Claude Code usage limits? Did they shrink? I’m hitting the limit within an hour on the Max plan.

I've been on macOS for 10 years, but month ago I was forced onto Windows 11. I had no clue what the hell I was doing, fucking hate this system. Tried to use Claude Code and have never touched a single goddamn setting since. So when something breaks, I just rip a YouTube transcript, throw it at Claude Code, and that bastard fixes it for me.

@dangreenheck on the other hand, you were satisfied with result even before, now it is even better! look cool

I made a dumb mistake with my water shader🤦🏻‍♂️ The wave evolution equation has two parts: a standing wave component and a traveling wave component. When I was first getting this all to work, I hard-coded constants so the waves were always traveling forward at full speed. By adding a "standing wave ratio" where I can control the ratio between the two terms, I can now achieve wave patterns that oscillate in place. The video on the left has SWR = 0 and the video on the right SWR = 0.7 (I exaggerated the waves a bit to make the difference more noticeable). Now I have to re-record all my marketing videos...😭

When AI hits something unpredictable in a big task, it doesn’t stop and say “this is broken, we need to fix the foundation first.” It just pushes through and produces garbage because something upstream is off. I’m trying to solve this at the planning level. On smaller, well-scoped tasks, it performs really well. Is that an issue of my planning / decomposition, or ai workflow I am not sure yet.

The most common way Claude Code fails is by making wrong assumptions. It assumes something and runs with it without properly validating, then ends up going down rabbit holes. How do you deal with that?

@dexhorthy gold

we've been trying a bunch of stuff. this one kinda works.

Claude: Let me just... Me: /clear

@shadcn What is the best model with visual understanding from your pov? I have been trying to apply agent-browser with claude to visual testing, no success.

To get translucent menus right, we had to manually test components over different gradients & colors, checking visuals & accessibility. Now multiply that by 5 component types across 5 styles in both light and dark modes. It's a lot of work. Work that, for now, AI can't do.

@athasdev I find that opening PRs when I am working alone a bit of overkill

@athasdev can i work locally, i.e. without PRs, main => worktree => rebase, but still have this nice sidebar per worktree?

Athas is a command center for agents.