This is an automatic post triggered within 2 minutes of an official Claude system status update. Incident: Claude Cowork not starting for some users Check on progress and whether or not the incident has been resolved yet here : https://sta…

Wanted Claude Desktop on my Arch setup so I made a PKGBUILD for it. It's a wrapper around aaddrick/claude-desktop-debian, which already does the main job of getting Claude running on Linux but I kind of hate app images and wanted a native…

So I already have a Mac Studio M4 Max (return window still available)with 64GB RAM, but I’m eyeing the Corsair AI Workstation 300 (Ryzen AI Max+ 395, 96 VRAM out of 128GB, $3,250). Both seem decent for running models locally with Ollama.

I've been collecting "jailbreak" and "unlock" prompts for 2 years. Most are either outdated, overhyped, or just wrong about how LLMs work.

Hi All, Not really sure if this is the correct place to post this, but it looks like the VS Code release mechanism might have been compromised. The Visual Studio Marketplace has a release today of 0.44.1, but the github page shows no such…

Inspired by Karpathy's autoresearch idea — an LLM runs training experiments autonomously to beat its own best score — but applied to code instead of ML training runs. I built this plugin as a way to set up an optimization loop on a codebas…

Completely unprompted btw . I was scared

We have been encouraged at work to use Co Pilot and to explore use cases for AI. As part of this exercise I started to (stupidly) use my personal claude account on my work computer to compare the quality of output.

King Louie An open-source, cross-platform AI chat desktop app. Bring your own API keys.

Introducing Timeplus AgentGuard, the first real-time security detection application purpose-built for AI agents. Running natively on the Timeplus engine, AgentGuard turns raw OpenTelemetry logs, metrics, traces plus agent hook events into…

Hello all just as it sounds. I recently started using GLM 5.1 in cursor 3 but unlike in the past, GLM 5.1 ran through my entire daily budget from summarizing chat context and running commands.

I am looking to create a benchmarking tool for LLM usage / pricing. My initial thought was that pricing in the space is quite opaque and people might want to see how their spend / pricing compares to other similar companies.

https://preview.redd.it/c0ns26pf3mvg1.png?width=1920&format=png&auto=webp&s=4b6ac114c2e0395684ac0ba79e591d71ccca2fe3 ParseBench lets you test the accuracy of different parsers using your own documents. Ran this across Gemini 3 flash, Qwen…

I need the simplest solution. I have an email account where clients contact me for help.

Claude Dispatch no longer creates a reply once its done with an assigned task, it only replies when the task is started to confirm that it has begun. I think this is because Dispatch now assigns its tasks to sub-agents in Cowork, so the ta…

The TheTom's turboquant's GPU accelerated turboquant (turbo3) has unlocked high context gains for the 35BA3B family. I can now achieve ~40tg/s via the following GPU-POOR compilation flags and configuration: cmake -B build -DGGML_CUDA=ON -D…

Shipped today. The benchmarks are real: 87.6% SWE-bench (from 80.8%), +13% on coding tasks, 3x more resolved production tasks on Rakuten-SWE-Bench.

{ "schema_version": "1.1", "protocol": "mcp", "protocol_version": "2025-03-26", "name": "HemmaBo Federation MCP Server", "description": "Direct booking infrastructure for vacation rentals. Search properties, check availability, get live pr…

We present SIR-Bench, a benchmark of 794 test cases for evaluating autonomous security incident response agents that distinguishes genuine forensic investigation from alert parroting. Derived from 129 anonymized incident patterns with expe…

Article Conversation Single-agent AI coding is a nightmare for engineers Created by and I pay my upfront subscription ($200/month), write what I hope is the right prompt (prompt AND context engineer), and wait. 35 minutes later, the agent…

Who's paying for tokens and why? (The Anthropic 1000) The information that would most clarify the nature of the AI boom right now is: who’s paying for tokens, and why?

Built a library to translate between OpenAI Responses/Assistants APIs and other provider APIs. Provides full compatibility so it’s a total drop-in regardless of which provider you use or which features (computer use, web search).

Recent work has shown that LLMs can sometimes detect when steering vectors are injected into their residual stream and identify the injected concept -- a phenomenon termed "introspective awareness." We investigate the mechanisms underlying…

Has anyone found a good way to manage model params based on the recommendations of the model developers that doesn't require manually managing a local config file? I have an ever growing bash script for launching llama.cpp server which inc…

Scaling autoregressive large language models (LLMs) has driven unprecedented progress but comes with vast computational costs. In this work, we tackle these costs by leveraging unstructured sparsity within an LLM's feedforward layers, the…

We've detected unusual activity from your computer network To continue, please click the box below to let us know you're not a robot. Why did this happen?

I built a Power BI workflow around Codex because I wanted something that could go beyond Microsoft's official powerbi-modeling-mcp. Their MCP handles semantic model operations well, but it stops short of local PBIR report authoring.

https://preview.redd.it/ttbzp6hexlvg1.png?width=995&format=png&auto=webp&s=4a65342507728c206b0b3a0f3e587d034489d4a1 While I was testing out Opus 4.7 on a highly complex Physics problem it told me it has "reached its max tokens to sample" a…

I've been trying to get back into coding but realistically I've got maybe 20-30 mins a day. Most tools either take forever to set up or feel like you need hours to get anything done Been looking into AI coding agents but not sure what actu…