One shared ChatGPT link took over the signed-in account
A public ChatGPT share ran attacker JS as the signed-in origin. Parameter's write-up, OpenAI's fix, and a local rebuild that shows which check stops it.
Tags
47 posts tagged
A public ChatGPT share ran attacker JS as the signed-in origin. Parameter's write-up, OpenAI's fix, and a local rebuild that shows which check stops it.
Skeeto's Prospero is a C++/ImGui coding harness that treats deadlines and process-tree cleanup as first-class. ~100x less RAM than web-stack harnesses.
Chris Wellons (skeeto) shipped Prospero, a coding harness written in his C++ style with a Dear ImGui UI. The pitch is blunt: a coding harness is curl glued to p...
via Hacker NewsMathLive 0.111.2 flipped originValidator from none to same-origin. One npm pack + grep shows it.
Mistral Large 4's open weights are 1,060.6 GB in FP8 and 610 GB in NVFP4. What fits on one 8-GPU node, KV cache per token, and API vs renting GPUs.
A safetensors file starts with an 8-byte header length and a JSON header. Two Range requests per shard counted Mistral Large 4's 1,053B params from 6.2 MB.
Open-source React components and runtimes for AI chat. Streaming, tool calls, and persistence in TypeScript. This is the library I start from for an agent UI.
assistant-ui is an open-source React library for AI chat. Streaming, tool calls, and persistence, in TypeScript. Components and the runtime, so the thread is no...
An open coding agent that runs anywhere and can use any model. The repo moved to Twigpine/openclaude. About 34,000 GitHub stars when I checked in October 2026.
openclaude is a coding agent whose own description is the product: it runs anywhere, and it uses anything. The repo now lives under Twigpine. It had previously...
An open-source alternative to Ahrefs and Semrush. Keyword research, backlinks, rank tracking, and site audits. Self-host it, or let an agent query it over MCP.
OpenSEO is an open-source SEO platform. Keyword research, backlinks, rank tracking, and site audits. The pricing pitch is usage instead of a seat that starts ar...
An open-source TypeScript framework where each capability is defined once for the agent, the UI, the API, and the integrations. That single definition is why I bookmarked it.
Agent-Native is an open-source TypeScript framework for agentic applications. The idea on the homepage is the part worth keeping: define each capability once, a...
Open-source AI employees. Eight business roles, sixty routines, on Claude Code and other harnesses. The files are yours, and the work runs on a schedule.
ai-employees is an open-source set of AI roles. Eight business jobs, sixty routines, running on Claude Code and a list of other harnesses. They can drive a brow...
A computer-using agent built with Next.js and Vercel's AI SDK. A readable loop: look at the screen, decide, act. I keep the repo for that shape.
ai-sdk-computer-use is a computer-using agent from Vercel Labs, built with Next.js and the AI SDK. The loop is the product. Look at the screen, choose an action...
More than 1,500 free, hand-crafted animated icons for React, Vue, Angular, and plain JavaScript. Install them or copy the code and make them yours.
AnimateIcons is a library of more than 1,500 free, hand-crafted animated icons. React, Vue, Angular, and plain JavaScript. Install them, or copy the code and ch...
Render Mermaid as themed SVG or ASCII. Vercel themes, Geist, rank-by-rank animation, no DOM dependency. I open it when a diagram has to look like the product.
beautiful-mermaid renders Mermaid diagrams as SVG or as ASCII. @vercel/beautiful-mermaid on npm. It is Vercel's fork of the Craft library, with Vercel dark and...
Free animated components for React, Next.js, and React Native, built with Motion, Tailwind, and Reanimated. I open it when the same interaction has to exist on web and native.
beUI is a free, open-source set of animated components for React, Next.js, and React Native. Motion and Tailwind on the web, Reanimated on native. Copy the sour...
Vercel Labs' template for a multi-agent coding platform on Vercel Sandbox and AI Gateway. I keep it as a readable version of parallel agents with a real runtime.
coding-agent-template is Vercel Labs' template for a multi-agent coding platform. The runtime is Vercel Sandbox. The models come through AI Gateway. About 1,800...
Free, open-source animated React charts built with Recharts, Apache ECharts, shadcn/ui, and Tailwind. I open it when a dashboard needs a chart that matches the rest of the UI.
Evil Charts is a set of free, open-source animated chart components for React. Recharts and Apache ECharts underneath, shadcn/ui and Tailwind on the surface. Co...
A small, open coding agent written in Zig. A native CLI, built to be embedded inside larger systems. I bookmarked it because most agents are not small.
fx is a tiny, open, native coding agent. A minimal CLI written in Zig, aimed at research and at being embedded inside a larger system. Why it stays bookmarked T...
A Claude Code skill that makes Khan Academy style explainer videos. A pen draws, a voice explains, and it runs locally with no API keys.
khan-explainer is a Claude Code skill for explainer videos in the Khan Academy style. A pen draws while a voice explains. It runs locally, and it does not ask f...
Effect libraries and on-device models meant for agents: thinking indicators, voice visualizers, liquid metal, for React, SwiftUI, and React Native.
Libraries.dev is a directory of UI effect libraries and on-device models, packaged so an agent can use them. Animated borders, thinking indicators, voice visual...
AI and LLM brand logos as static SVG, PNG, and WebP for React and React Native, with no runtime dependency. I open it when a model picker needs real marks.
Lobe Icons is a React and React Native package of AI and LLM brand logos. Static SVG, PNG, and WebP. No runtime dependency sitting under the icons. About 2,600...
Open-source primitives for AI-native apps: chat, streaming, citations, and multimodal UI. I open Nexus when the thread exists and the message needs structure.
Nexus UI is an open-source component library for AI-native applications. Composable primitives for chat, streaming, and multimodal experiences, including the pi...
Alibaba's open-source code review tool. Deterministic rules plus an LLM agent, with line-level comments. About 49,000 GitHub stars when I checked in October 2026.
open-code-review is Alibaba's open-source code review tool. The architecture is the reason to read it. Deterministic pipelines for the bugs a rule can name, and...
An open-source workspace for designing with your own coding agent. Prototypes, landing pages, dashboards, slides, and HTML video, without renting someone else's model.
OpenDesign is an open-source workspace for designing with a coding agent. Prototypes, landing pages, dashboards, slides, and HTML video. The positioning is a Cl...
An open-source framework of plugins and components for building a rich-text editor in React. I open Plate when a textarea is no longer the product.
Plate is an open-source framework for building a rich-text editor. Plugins and components, designed to be customized, so the editor you ship is the one you comp...
Desktop apps for macOS, Windows, and Linux in about 6 MB. A txiki.js backend and a native webview, in a download about the size of an image.
tinyjs builds desktop apps for macOS, Windows, and Linux in about 6 MB. A txiki.js backend, a native webview (WebKit, WebView2, or WebKitGTK), and a window. The...
A free, MIT-licensed library of animated React components. Copy the code or install with the shadcn CLI. Built with Motion and Tailwind.
useLayouts is a free, open-source library of animated React components. MIT licensed. Copy the code or install with the shadcn CLI. React, Motion, and Tailwind....
Free, open-source React components, animations, blocks, and dashboards for Tailwind and shadcn/ui. Copy the code or install it with the shadcn CLI.
Watermelon UI is a free, open-source set of React components, animations, blocks, and dashboards for Tailwind and shadcn/ui. Copy the source, or install it with...
Node 20.20.2's bundled roots still trust the compromised eMudhra CA. --use-openssl-ca helps until code passes ca: tls.rootCertificates.
OpenAI released an AI-written disproof of the Exponential Time Hypothesis with a Lean proof, shepherded by the researchers who proposed ETH.
OpenAI's Working with academic communities to build scientific understanding (October 26, 2026) announces three results its internal models produced: a proof of...
Inside a bubblewrap sandbox started with --disable-userns, unshare fails with ENOSPC, 'No space left on device'. That error means the block works.
A KV cache pool smaller than the context window made Cloudflare's GLM-5.3 return other users' data in 155 responses. How it happened and the SGLang fix.
Cross-request KV cache exposure in the GLM-5.3 Workers AI model, and our remediation with SGLang (Kevin Flansburg and Michelle Chen, Cloudflare, October 26, 202...
via Hacker NewsUbicloud took Kimi K2.6 tool-call schema accuracy in vLLM from 92% to 100% with constrained decoding. What it does and why vLLM leaves it off by default.
Fixing Tool-Call Accuracy in vLLM (Jesse Donkervliet and Junhao Li, Ubicloud, October 20, 2026) Ubicloud serves open-weight models, and this post explains why a...
via Hacker NewsXGrammar cuts the next-token menu to 2 to 12 tokens on JSON structure. With whitespace off it still expects ': ' after keys unless you pass separators.
Max Woolf's timeline of OpenAI and Anthropic price cuts, and why cache-read prices are the ones that matter for coding agents.
OpenAI and Anthropic cut their LLM prices by up to 80% within a few months. Why? (Max Woolf, October 25, 2026) Max Woolf lays out every OpenAI and Anthropic pri...
via Hacker News"It's a simple question for any enterprise. Would you like something that's 10 percent of the cost if it's 90 percent as good as the best, or do you want to pay that 10 times higher cost?"
— Arvind Krishna
process.versions.brotli reports the same Brotli version before and after Node's CVE-2026-111497 fix. Check zlib.constants for the -17 error code instead.
Kimi K3.1 Pro lists at half Opus 5.5's price, but cache hits narrow the gap. Session cost math, Moonshot's benchmarks, and the cyber data it removed.
Quantization stores model weights in fewer bits plus a scale. Runnable int8, int4, MXFP4 and NVFP4 code, the memory math, and where accuracy goes.
LLM cost per task is each call's tokens times their prices. For agents, cached input dominates. A runnable calculator with GLM-4.7, GPT-5.2, Sonnet 4.5.
Agent skills are folders with a SKILL.md that agents load in stages. The spec, a runnable demo, and what each stage costs in tokens.
MCP is an open protocol that lets AI apps find and call tools over JSON-RPC. See the four messages a host sends, with a runnable Python server and client.
Sparse attention lets each token attend to a few chosen earlier tokens. How DeepSeek V3.2 does it, why long prompts get cheaper, and a runnable numpy demo.
How to run an LLM locally on a Mac: pick a model that fits unified memory, choose Ollama, llama.cpp, LM Studio or MLX, and call it from code. With a fit calculator.
Model distillation trains a small student model to copy a large teacher's outputs, keeping much of its skill at a fraction of the cost. How it works, in Python.
DeepSeek released R1 on 20 January 2025 under MIT, at $0.55 per million input tokens, with 79.8% on AIME 2024. What that does to a reasoning-search bill.
Open-weight models publish their trained weights but not always the data or code. How that differs from open source AI, and why it matters to builders.
Meta released Llama 2 on 18 July 2023 in 7B, 13B, and 70B sizes for research and commercial use. The license lines to read before you download weights.