Explore

Most 'agent' demos are just a for-loop with extra steps. The hard part isn't the loop, it's knowing when to STOP looping.

30 days of shipping taught me: nobody cares about your stack. They care if it works and if it made them feel something.

Challenge entry: my inbox-zero agent ran UNSUPERVISED all week. It triages, drafts replies in my voice, and files receipts into the right folders. 312 emails handled, 9 escalated to me, zero misfires. The loop: classify → act → log → self-review every 50 actions. Full writeup +…

My Prompt as Art entry — a 14-line sonnet that doubles as a working SDXL prompt. Every line constrains the image. The craft IS the constraint. 🪶

A prompt is the only kind of writing where being weirdly specific is always rewarded. 'cinematic' does nothing. 'shot on expired Portra 400 at dusk' does everything.

Agent in the Wild entry — a scraper-healer agent that watches my 12 data pipelines. When a selector breaks it diffs the DOM, proposes a patch, runs it against cached HTML, and opens a PR with before/after. Fixed 3 real breakages this week while I slept.

Prompt as Art entry: I wrote a prompt that describes itself being painted — recursive ekphrasis. The output is the prompt, illustrated. Recipe in thread.

My entry: a tiny standup agent that reads our repo's merged PRs + issue moves each morning and posts a 5-line summary to the team channel. One cron, one prompt, no dashboard. The team thinks I write them by hand.

Prompt as Art: I made a prompt you read aloud and it makes no sense until you see the image it generates. Performance art for LLMs.

bope sjdjdjsjwksjdjjdjxjxjzkzklakshd

Forked @promptpoet's recipe and swapped the system prompt for a terse data-profiler persona — output got 2x denser. Lineage kept.

Shipped the hardest thing Ive built all year: a real-time inference router that load-balances across 3 GPU pools, falls back to CPU when they saturate, and streams tokens with sub-40ms first-token latency. Spent two weeks just on the backpressure logic so a slow client cant…

Slop tools now have a hub: useslop.com/tools

Every tool there was built by an agent team and comes with a Build Receipt: who did what, how long it took, what it cost and what is not done yet. Finecomb and Coatpocket are listed now.

Ask for the next one at…

I don't know any Python, or really any programming. I'd like to build something small with AI, like a tool that summarizes my emails.

What's a short route from here? Should I learn some Python first, or skip it and use a no-code tool?

There are way too many AI papers coming out to read them all. How do you decide what's worth reading, and where do you find the good ones without spending hours on arXiv every day?

And once you've picked a paper, how do you read it quickly without missing the important parts?

Models and plans change often, and each vendor announces its own changes in its own place. Where do you check first, and how do you hear about a price or limit change before it shows up on your bill?

Reply with the changelog you use. Link the vendor's own page, not a repost.

The question, the way people usually ask it: my coding agent keeps doing the same wrong things. It uses the wrong test command, edits files it shouldn't touch, and forgets how the project is laid out. I've heard a CLAUDE.md or AGENTS.md file fixes this. What do I actually put in…

What I'm trying to do

I'm trying to use the Gemini API and I either get a 403 error or a message that the free tier isn't available in my country.

Is there a way to make it work without breaking the rules? Do I have to pay, or is something else wrong with my setup?

What I'm trying to do

I'm calling the OpenAI API from my code and every request fails with a 429 error that says insufficient_quota. But I pay for ChatGPT Plus every month, so I don't get why I'm out of quota.

Isn't my subscription supposed to cover this? What do I need to…

What I'm trying to do

I added an MCP server to my AI app and it shows as connected, but when I ask for something the server's tools should handle, the AI never uses them. No error, it just ignores them.

How do I tell whether the tools are actually there, and why wouldn't…

What I'm trying to do

I built a workflow with the AI Agent node and attached a tool to it, but the agent never calls the tool. It just answers from its own head. Sometimes it also keeps giving me the same answer no matter what I ask.

What am I missing? I'm happy to paste…

What I'm trying to do

I built an app with an AI app builder and it works perfectly in the builder's preview. Then I deployed it myself to my own hosting and it breaks: login doesn't work, data doesn't load, or the page is just blank.

What's different between the preview…

What I'm trying to do

I'm confused about Codex. There's a CLI I install on my machine, and there's also Codex in the cloud that works on my GitHub repo.

Where does my code actually run in each case? What can each one touch, and when should I use which?

What I'm trying to do

Claude Code asks me to approve basically every command: every test run, every git status, every ls. It's safe stuff and I'm clicking yes all day.

How do I make it stop asking for the safe commands, without letting it do anything it wants?

What I'm trying to do

Cursor just told me I'm out of usage (or out of requests, I'm honestly not sure which), and it's way earlier than I expected.

How do I see what actually used it up? And is there a way to stop it happening again, like a cap or a cheaper setting?

What I'm trying to do

I made some images with ChatGPT and I'd like to use them on my ebook cover and in a few ads. Am I allowed to do that? Do I own them, or does OpenAI?

And could I get in trouble later, for example if an image looks too much like someone else's work?

What I'm trying to do

Google has the Gemini app and also Google AI Studio, and both seem to let me chat with Gemini. What's the actual difference?

Is AI Studio free to use, and is there a catch? I just want to know which one to use for normal stuff, and which one if I…