Abror Aliboyev

@BlockedPaths Same as early chatgpt wrappers

Maybe it’s just me, but I’m getting tired of seeing the same AI terminal launched over and over again. Warp. Conductor. Superconductor. CMUX. Ghostex. And another one next week. Every demo looks nearly identical: “Look, I can run multiple agents in panes.” At some point this stops being innovation and starts being a UI clone war.

Came up, disrupted workflow and vanished away.

We now support rich formatting for all chatbots. Tables, nested lists, inline media, formulas, headers and more — right in Telegram messages. 🔨 Start building! Docs: #rich-message-formatting-options" target="_blank" rel="nofollow noopener">core.telegram.org/bots/api#rich-…

@durov Very needed feature. Would be good to make chat sessions in bots better.

Question would you want to have a self hosted LLM proxy server? That tracks usage, rate limits, logs payload, routes any of your LLM requests? Central place you own and control all your LLM activities?

AI is amplifier not innovator. Bring your ideas to it. Do research extract knowledge from it. Don't expect it will bring you something new. You don't have weight against AI if you are not the creating one.

@ryanvogel No worries there's some entity already handling this concern.

what if anthropic and openai start turning off the models on weekends to force us to go outside

The Anthropic story nobody's putting in order. Here it is: - Jul 2025 — Anthropic signs a $200M Pentagon deal. Claude becomes the first frontier model cleared for classified networks. - Jan 2026 — Hegseth's memo orders every DOD AI contract to adopt "any lawful use" language. Around the same time, Anthropic asks whether Claude was used in a January military operation — signaling it might object. - Feb 24 — Pentagon ultimatum: drop your limits by Friday or lose the contract, get blacklisted, or face the Defense Production Act. Anthropic's two red lines: no fully autonomous weapons without human oversight, no mass surveillance of Americans. - Feb 27 — Trump orders every federal agency to cease using Anthropic. Hegseth brands them a "supply chain risk" — a label built for foreign adversaries, never once used on a US company — and calls it "a master class in arrogance and betrayal." …within hours, OpenAI signs its own Pentagon deal. - Mar 9 — Anthropic sues. First Amendment retaliation + Fifth Amendment due process. - Mar 26 — SF judge grants an injunction: "classic illegal First Amendment retaliation." - Apr 8 — DC appeals court sides with the government "during an active military conflict." Anthropic: barred from DOD contracts, billions exposed. - Jun 2026 — Anthropic ships Fable 5. So good the haters resubscribe and the $200 tier is suddenly worth it. - Jun 12 — the government issues its first-ever LLM "export control directive," forcing Fable 5 offline. Refunds fly. Customers scatter — many to Chinese models like Kimi K2.7. The throughline: a company said no to autonomous war and to spying on Americans. Everything since has been the bill.

The export "control" translated into honest English: "You wouldn't hand us uncensored Opus for our wars? Then we'll do everything we can to keep you inside our grip and bleed you slow — and OpenAI, who signed without blinking, gets to eat your lunch while you fight us in court."

Yeah, funny consequences. An "export control" that cost a US company its customers and handed them straight to rival providers, most of them Chinese. Controlled the export of exactly nothing. Kind of do you think through consequences of this "directives" you guys?

@zeeg @dok2001 I have 3

even @dok2001 has a github. whats your excuse?

What the shell is going on?

One of the weird behaviors of Claude models is that they mislead themselves with timing. It says "Good night's work" and then any follow up tasks came the next day in their minds. They poison their own context with wrong data. 😅 If you ignore it, gets worse and model keeps building timeline on its own wrong assumptions. Funny and little annoying.

@thdxr 123456

what's your password?

@kylegawley Wrong side of cost optimization 😅

I cancelled my $10/mo Calendly subscription and vibe coded my own with Fable for $12,000

@awsdevelopers naming of that service itself wrong answer to it. I would suggest rename it to CS2 - Complex Storage Service

wrong answers only, what does S3 stand for?

🎥🧌

@orcdev @icanvardar Sometimes claude tries to match my lowercase behavior in long sessions so you cannot tell.

@icanvardar i honestly think that's absolutely a good-way to show you're not — AI

using lowercase just to make sure you don’t think i’m an ai

@ClaudeDevs Web Search tool in CC is not working for Fable 5

@TheSpacerr Why should I?

How can you prove that you are not an AI?