On mass-produced math: Reviews the verification status and transparency gaps of OpenAI’s batch of mathematical preprints, and their implications for mathematical research.

OpenAI Math Dump Hits My Home: Hung Le documents his ongoing reading of graph-metric embedding manuscripts, criticizing their compressed exposition while sharing what he is learning from the methods.

TCS, AI, and the Adults in the Room: Clément Canonne urges senior theoretical computer scientists to articulate their field’s values and secure funding and stable positions for junior researchers.

Interview with Michael I. Jordan: Michael I. Jordan discusses AI’s social role and the effects of poorly grounded extinction-risk narratives on young people; his accompanying note also cautions against overreading the headline.

Jelani Nelson on joining Anthropic: Jelani Nelson describes how trusted researchers’ views on model capabilities led him to reconsider his expectations and, after learning about several companies, join Anthropic. He also worries that society is unprepared for changes to intellectual work. This is his personal account and assessment.

Simons mathematics and AI initiative: The Simons Foundation, Anthropic, and OpenAI commit a combined $15 million in funding and resources to help mathematicians evaluate AI-assisted results, learn tools, and shape shared standards.

Thinking will become a hobby: Diomidis Spinellis discusses AI’s effects on cognitive work and how people might build meaningful lives beyond work.

Gowers on AI extinction risk: Timothy Gowers examines assumptions and uncertainties in arguments about AI extinction risk and explains why mathematicians should study them seriously.

Neural Networks Are Already Clean: Nick Cammarata proposes a working hypothesis: some apparently messy neurons may reflect limitations in tools, explanations, or training, and network mechanisms may be easier to understand than expected.

The Beauty of Something Grown: Dean W. Ball considers how AI safety and policy debates might change if Goodfire’s interpretability hypothesis proves correct.

Expectations at the frontier: Anthropic combines internal employment observations with a public survey to explore associations between experience with AI and expectations for work and the economy.

Models / Agents / Scientific Applications

Muse Wonder: Meta presents capabilities for personal and professional agents, including cross-conversation memory, sustained task execution, and coding.

Muse Spark Open: Meta releases model weights for multimodal coding and long-horizon tasks, with a local demonstration across two M5 Ultra Mac Studios. The weights use the Muse Model License.

Span-02: Respan introduces an open-weight model for decision tasks, with time-budgeted reasoning that can stop and return its best answer so far.

Frontier pre-training at Microsoft AI: Details the pre-training of MAI-Base-2, covering scaling science, infrastructure, and data pipelines for its mixture-of-experts model.

Gemini 4 Argon for Robotics: Google demonstrates robot control, real-to-simulation reconstruction, video annotation, and iterative improvement.

Research@ 2026: Google rounds up ten research developments spanning geospatial analysis, research assistance, technical-document review, numerical forecasting, and health AI.

Heaviside and Microsoft Discovery: Arena Physica brings its electromagnetic field model to Microsoft Discovery to assist antenna, photonics, and semiconductor design with fast predictions, while retaining traditional solvers for final verification.

Quantum circuit optimization with Claude: Anthropic reports optimizing eight quantum circuits with automated evaluation, substantially reducing estimated resource requirements on some tasks.

Autonomous wet labs: Google presents research connecting AI agents, lab robots, and experimental feedback through laboratory digital twins, aptamer screening, and protein-binder design.

Origins of ECDSA curves: Anthropic reports that Claude reconstructed how several NIST elliptic-curve seeds were generated, providing reproducible code to address questions about their provenance.

Defenses against distillation attacks: Anthropic outlines defenses against unauthorized model distillation, reporting early reductions of roughly 90–99% in harvested data while noting uncertainty about durability and effects on legitimate requests.

Small-model self-replication evaluations: Uses IoT inference experiments and simplified replication evaluations to study the prospects and barriers for small models to replicate autonomously on ordinary connected hardware.

Scion source control: Anthropic introduces its internal source-control system and explains how AI-generated code changes storage, checkout, and integration needs in a very large monorepo.

TheoPhysEval: Evaluates autonomous AI agents on 50 verifiable research-level theoretical physics problems.

Auto Research: Tracks how community pull requests improve key metrics in OpenAI’s 2026 mathematical preprints.

Mathematics / Research / Learning

Lean 4 teacher’s guide for mathematics: A teacher’s guide that introduces formal proofs to mathematics students without prior programming experience through eight topics.

Acorn: A theorem prover for writing and checking formal proofs.

Prove2Me: A collaborative Lean 4 platform that turns results from papers or textbooks into open proof missions and publishes proved results to Formalpedia.

RL Handbook: A handbook on reinforcement learning and sequential decision making, covering bandits, Markov decision processes, policy gradients, offline learning, and multi-agent methods.

Stochastic Superoptimization: Formulates loop-free binary superoptimization as stochastic search, using correctness and performance costs to guide STOKE.

Probability distribution map: An interactive map of common distributions, their properties and uses, and relationships through special cases, limits, and transformations.

Deep Learning Alchemy: A Stanford deep learning course centered on experiments, predicting their outcomes, and understanding empirical phenomena in neural networks.

Principles and Practice of Deep Representation Learning: A Chinese textbook on the principles and practice of deep representation learning.

Why is Fusion Hard?: Traces the difficulty of fusion through plasma confinement and instabilities, explaining how successive physical obstacles arise.

Introduction to Cosmology: Cosmology lecture notes focusing on large-scale structure, cosmic microwave background anisotropy, and inflation.

Concise Research Roadmap (PDF): Jared Kaplan’s research roadmap, updated in July 2021, covering general AI assistants, scaling, and interpretability.

Training performance and activation checkpoint offloading: An ML engineering performance guide, including a section on offloading activation checkpoints to CPU memory to reduce GPU memory use.

Quant Wiki: An open-source project making quantitative finance knowledge accessible in Chinese.

LLMQuant: A community focused on AI, large language models, and quantitative finance, with projects including a quant wiki and paper discovery.

Programming / Agent Tools

OpenIntelligentUI: An open-source generative UI framework that turns chat responses into explanations, comparisons, and interactive tools.

OpenUI: Open-source infrastructure for AI apps and agents to generate interactive interfaces using existing components and design systems.

REA: An agent-based reverse-engineering tool for examining app behavior and native binaries.

open-claude-ui: A project recreating the Claude desktop interface through reverse engineering.

Mirror: Reflection’s coding agent harness.

Earshot: An open-source, CPU-only tool for coding agents to analyze, review, and create sound.

Yomado: A virtual Mac screen that lets visual agents operate desktop apps through screenshots and input.

Codex++ 2.0.0: The new release of a tweak system for the Codex desktop app.

Mayoi: A lightweight code editor written in Go with a native GPU-rendered interface.

ELYTerminal: A terminal emulator built with GPUI and Ely.

MathJax Go: A native Go port of the MathJax 3.2.2 TeX-to-SVG pipeline used by D2.

ManimGX: An animation engine for math and 3D videos, with Manim’s Python API and a Rust GPU core.

Pele: A lightweight open-source library that renders Mermaid diagrams as SVG themed with CSS variables.

Design / Writing / Desktop

Anatomy: A Claude Code skill that turns concepts into interactive isometric mechanical illustrations.

Guizang PPT Skill: An agent skill for generating HTML slide decks, with layout guidance, image prompts, and a presentation runtime.

awesome-design-md: A collection of DESIGN.md files describing popular brand design systems for coding agents.

anti-defensive-writing: A Codex writing skill for removing defensive phrasing and making prose more direct.

UX Writing Skill: An agent skill for Claude and Codex that applies design system rules to interface copy.

Longlist: A directory of design studios and independent designers for finding collaborators or employers.

Build your slide template: Explains how to build a simple, reusable slide template from a color palette, typography, and blank layouts.

MacSpace: Inspects and cleans macOS System Data and Apple Intelligence model files.

OpenClean: A Mac cleaner for removing junk, fully uninstalling apps, and inspecting disk usage.

Inua: A desktop shell for Hyprland with a bar, launcher, control center, notifications, dock, and settings.

POND: An Arch Linux-based distribution using Niri with a curated desktop and system configuration.

Craft Launcher: A launcher for installing, updating, and opening apps in the Craft App Suite.

v2mac: A native macOS client for Xray-core with proxy subscriptions, latency tests, routing, and menu bar controls.

UU Remote Linux host: An unofficial Linux host implementation for NetEase UU Remote.

UU Remote integration guide: Documents the URL Scheme for device pairing and remote control, with iOS and Android examples and error codes.

Epic Games Launcher: Epic Games’ desktop launcher for accessing and launching games.

Diachronic: Explores human behavior modeling and evaluation with Kairos 1.

I Quit OpenAI Because Its Culture Is Broken | The Atlantic: David Robinson explains why he left OpenAI, criticizes its safety culture, and argues for learning from other industries that manage serious risks.

Introducing Beam | Reflection: Reflection introduces Beam, a 501-billion-parameter mixture-of-experts model for coding, reasoning, and agent tasks, with weights planned for release later this month.

Rho-1: Collapsing the multimodal stack | Reka: Reka previews Rho-1, which combines text, images, video, and actions in a single model.

Explainability as a form of system closure: Frames explanation as building approximately closed models over variables humans can grasp, using chaotic systems to explore the limits of interpretability.

Introducing Reason Agent 1.0 · Reason Agent: Reason Machines introduces Reason Agent 1.0 for autonomous software engineering.

Introducing TasteVal | P-Zero Research · TasteVal: Measuring the Experimental Research Taste of AI Systems Against Human Experts: Evaluates how models design experiments and interpret results on fixed research problems, comparing experimental compute efficiency with human experts.

Mathematics / Research / Learning

Practical Parametric Yield | Irrational Analysis: Explains chip parametric yield through process variation, clock speed, and power, then discusses the tradeoffs facing Cerebras’s wafer-scale computing.

A Riemann Hypothesis Explainer Made with Opus 5.5: The uploader presents a Riemann hypothesis explainer with video made using Claude Opus 5.5 and narration provided by Gemini.

Two Room-Temperature Antiferromagnetic Semiconductor Candidates | Vals AI: Reports two candidate room-temperature antiferromagnetic semiconductors found with Claude agents, sharing calculations and limitations; their properties still need experimental validation.

Presenting a proof of the Kannan-Lovasz-Simonovits conjecture: Bizeul, Klartag, and Lehec present a 32-page manuscript proposing a complete proof of the KLS conjecture, disclosing that ChatGPT supplied most of the proof and mathematical ideas.

An $O(1)$ Bound for the KLS Constant: Song and Zhang propose an $O(1)$ upper bound for the KLS constant in the second version of their preprint, claiming to resolve the full conjecture.

A Dimension-Free Bound on the Poincaré Constant of Isotropic Log-Concave Measures: Balasubramanian and Kasiviswanathan release another 53-page manuscript claiming to prove the KLS conjecture, also disclosing AI involvement.

Stark units for real quadratic fields and reciprocity laws: Proposes reciprocity laws for Stark units in real quadratic fields, claiming to answer this case of Hilbert’s twelfth problem and derive Zauner’s existence conjecture in every dimension.

The Squares Project · n = 11 Case: Collects constructions, lower bounds, and machine verification records for packing eleven unit squares into a square; the project reports matching bounds, with its review record still pending.

The Fibonacci Geometry of Eleven Squares (PDF): Describes Trump’s eleven-square packing using flat tori and Fibonacci geometry, analyzing the unique minimum within its two-parameter contact family.

Sharing AI progress in mathematics | OpenAI · openai/math: OpenAI releases manuscripts, proofs, and verification materials from AI-assisted mathematical research, noting that the results still require review by the mathematical community.

Beyond n log n · Integer multiplication: Tracks conditional improvements to integer multiplication complexity bounds and related verification materials.

CrocSwap/integer-mult-bounds: Investigates conditional improvements to the exponent saving in OpenAI’s integer multiplication result #109.

The Mathocalypse | Scott Aaronson: Scott Aaronson discusses recent AI mathematics results and their implications for research, verification, and mathematicians.

What does an improved upper bound on $\omega$ mean | Abbas Mehrabian: Explains what improving the upper bound on the matrix multiplication exponent $\omega$ means.

On Matrix Multiplication Algorithms · Fast Matrix Multiplication (PDF): A lecture and survey on matrix multiplication algorithms, covering fast algorithms and complexity research.

Erich Friedman · Erich’s Packing Center · Cubes in Cubes: Erich Friedman’s mathematical puzzles and geometric packing resources, including records for packing cubes into a cube.

Nopert Polyhedra Explorer: Interactively explores polyhedra in Rupert’s problem through rotation, cutting holes, and passage tests.

Relativity Park: Walking Speed of Light: Sets the speed of light to walking speed so visitors can observe relativistic effects in a park scene.

Soft-to-Rigid Packing: Compares packing searches that gradually harden soft shapes with other methods, sharing experiment logs, code, replays, and their respective strengths and limitations.

News | Yasuyuki Kawahigashi: Yasuyuki Kawahigashi records operator algebra researchers’ reactions to OpenAI manuscripts and progress toward proofs humans can read.

A Brief Digest of Some New OpenAI Results | Persiflage: Frank Calegari comments on some newly released OpenAI mathematics results.

Big O Speedrun: Tracks complexity bounds for integer multiplication, Fourier transforms, 3SUM, all-pairs shortest paths, and other problems, distinguishing claims from verification status.

A reaction to the reactions | Balázs Pozsgay: Responds to discussions about AI mathematics from a theoretical physicist’s perspective and reflects on becoming a student again.

Programming / Systems / AI Tools

egoist/godiff: A native Git diff viewer for reviewing working trees, commits, and GitHub pull requests.

sahenjp/rustdsh: A Rust launcher for DeepSeek Harness with configuration migration and local tools.

How fast is Python 3.15? | Miguel Grinberg: Compares the Python 3.15 release candidate with older interpreters using recursive and sorting programs, including just-in-time compilation and free-threading modes.

sshx: Shares terminals live in the browser for collaboration.

Public testing is beginning | Superlogical: Rex begins inviting public testers, initially on macOS, with persistent sessions and remote connections.

egoist/gorex: A Rex terminal clone built with Go and MyGo’s native interface.

ArtCraft Crafting Apps · GitHub · VectorCraft · WordCraft · DeckCraft · CADCraft: A collection of open-source creative applications built in Rust for vector graphics, documents, presentations, and computer-aided design.

palemoky/chinese-poetry-api: A project providing Chinese poetry search and an API service.

Freaction/Aquilum: A local Markdown knowledge base for Windows with live preview, search, a relationship graph, a reader, and Model Context Protocol support.

solidsf-inc/sfFFT: An MIT-licensed fused Tensor Core fast Fourier transform convolution kernel for NVIDIA GB10.

Ely Office · ZacharyZhang-NY/ElyOffice: The Ely Office website and its Rust / GPUI desktop office suite project link; the repository currently returns 404.

RelayKit | New API: A Go module that converts requests, responses, and streaming events among four text protocols from OpenAI, Anthropic, and Gemini.

Tern | Stencil: A native Rust terminal and multiplexer compatible with Kitty, offering persistent sessions and cross-device connections; currently in closed testing.

There are many themes, but this one is yours | Earendil: Explains how Pi reads terminal colors to generate its system theme, discussing palettes and interface contrast.

Rewriting Prime Agent in Rust: Prime Intellect shares how multiple agents rewrote Prime Agent from TypeScript into Rust and how the team verified the rewrite.

Tov: An experimental language for coding agents with TypeScript-style syntax, native binary compilation, and structured compiler errors with suggested fixes.

Design / macOS / Other Finds

RevPDF: A PDF tool for offline editing, comparison, and format conversion.

An Opus 5.5 Video Recreation | R: Shares an Opus 5.5 creation in response to a video recreation challenge.

Paper Mono: A free, open-source monospace font with variable weights, coding ligatures, and a glyph showcase.

celve/uvim: A menu-bar application that brings Vim’s modal editing to macOS text fields.

AIME: An open-source Chinese input method for macOS built on RIME, with dictionary updates, saved phrases, and translation and rewriting of selected text.

Designeer: A curated directory of interface design tools, components, and inspiration.

alejandrobujan/tendedero: A native macOS app that hangs screenshots on a line along the top of the screen for temporary reference.

Pulse: Displays processes, processor usage, memory, network, energy use, and other system metrics in one native window.

Nongfsq/ftop: A floating system monitor for Apple Silicon Macs showing processor, memory, network, and process information.

Live Training Run · BrushArena: BrushArena’s live training page.

Introducing Julia 1 | Supersonic Labs: Supersonic Labs’ official introduction to Julia 1.

Introducing Claude Sonnet 5.5 | Anthropic: Anthropic introduces Sonnet 5.5’s improvements in coding, knowledge work, and running costs.

Introducing GPT-6.1 Sol | OpenAI: OpenAI introduces GPT-6.1 Sol and its updates in coding, computer use, and professional tasks.

Exploring Similarity Search with Prompted Subspaces — Jasper Lu: Uses text prompts to construct concept subspaces for image similarity search by color, style, or meaning.

RL is an evolutionary algorithm: Discusses reinforcement learning through the lens of evolutionary algorithms.

Introducing Gemini 4: Google introduces Gemini 4 Argon with an output limit of one million tokens, initially rolling out to trusted cyber defenders and testers.

ExplorationBench: Evaluates autonomous exploration in unfamiliar programming and logic environments, using interpreters and proof checkers to verify results.

Grokipedia: The Grokipedia homepage.

Cogentic: The Cogentic project homepage.

How many AI agents could we run? | Epoch AI: Estimates that high-bandwidth memory shipped in 2025–2027, fully deployed and allocated to the relevant workloads, could support nonstop agents providing weekly working hours equivalent to roughly 140–720 million full-time employees.

AI chips shipped through 2027 could run tens to hundreds of millions of concurrent frontier-model agents. Running nonstop, these agents would supply as many weekly working hours as about 140–720 million full-time employees.

A model guide for the GPT-6 family | OpenAI: A guide to model selection, instructions, context and cost management, long tasks, and production deployment with the GPT-6 family.

Lessons from building an automated research scaffold — LessWrong: Shares lessons from an automated research scaffold: limited productivity gains, alongside failures involving reward gaming, result interpretation, and audit monitoring.

What if automating AI R&D triggers an intelligence explosion? — CASP: Examines whether automated AI research could trigger an intelligence explosion, its implications, and policy responses, while emphasizing uncertainty.

Tokenization: A Survey for Modern NLP | alphaXiv · PDF: A survey of subword tokenization methods, their limitations, and alternatives in modern natural language processing.

Language Model “Shape” | Alex L. Zhang: Explores designing language-model input and output structures around agent scaffolds.

Mathematics / Research / Learning

synthetic-sciences/openscience: An open-source AI workbench for scientific research.

CSAPP Chinese Reading: A Chinese reading site for Computer Systems: A Programmer’s Perspective.

AGMAI: September 29 Statement: The mathematics advisory group draws on community feedback to recommend practices for publishing AI-generated mathematical results and supporting human understanding and review.

Transformer Explainer: An interactive visualization of GPT-2 small that explains computations inside a Transformer.

AI solves a ‘holy grail’ problem from probability theory | Scientific American: Reports on an Anthropic model’s proposed proof of a percolation continuity conjecture and discusses its implications for mathematics.

Apex Leaderboard: The Apex leaderboard announcement page.

MIT 6.5940 Fall 2026 TinyML and Efficient AI Computing: MIT’s fall 2026 course covering pruning, quantization, model serving, distributed training, and deployment on edge devices.

Expenditure Horizon | METR: Measures optimization ability through the intersection of human and agent cost–benefit curves, with an initial study of NanoGPT optimization.

Solving Open Research Problems Together | Meta AI Research: Describes Muse Spark’s collaboration with mathematicians on open research problems, with results checked by independent mathematicians.

Learning Forever using Artificial Neural Networks: Shibhansh Dohare’s doctoral thesis on loss of plasticity, continual backpropagation, and policy collapse.

Programming / Systems / AI Tools

A shared cache for Cargo builds | mr boxington: A shared Cargo build cache that reduces repeated compilation.

deepseek-ai/DeepGEMM-Ascend: A matrix multiplication kernel library for Huawei Ascend processors.

Introducing cf | Cloudflare: Cloudflare introduces an agent-oriented command-line tool for its full API.

Ely · GPUI components: A showcase of GPUI components.

claude.dev: A technical blog for developers building with Claude.

MonoCode: A graphical interface for multiple coding agents.

MyGo: A Go framework for desktop apps with web frontends or native interfaces.

Pi 1.0 | Earendil: Introduces Pi 1.0, an extensible agent framework.

Comma: A continuously running agent for personal tasks.

Lorca: Runs AI teammates locally to work on files and tasks through individual or group chats.

I’m Using GPUI For Everything | Freek Van der Herten: Recommends a Syntax team video about developing with GPUI.

Design / macOS / Other Finds

mg-chao/snow-apps: The source repository for Snow Shot and Snow Image Viewer.

Cida: Translates and polishes text across Mac apps.

tornikegomareli/omarchy-spaces: Shows app icons for each workspace in the Omarchy bar and supports workspace switching.

Pal-AI-Lab/Coopanion: Coo is a desktop companion at the bottom of the screen, supporting chat, voice, and authorized computer use.

The Making of Cursor’s Icons | Minor Adventures: Shares the design process behind Cursor’s icons.

IPPure: Checks IP information, risk, network egress, and leaks.

arXiv: Updated Submission Rate-Limit Policy: arXiv’s official announcement of its updated submission rate-limit policy.

adileo/squirreldisk: A cross-platform disk usage analyzer written in Rust, with sunburst and treemap views.

Yab: yet another browser: The Yab browser homepage.

AI Haotan: Compares prices and stock across marketplaces for AI accounts.

metasequoiaime/msime-windows: The Windows repository for Metasequoia IME, including its input framework, server, interface, and installer.

gillesgoetsch/omacvm: Runs Omarchy in a Mac virtual machine, integrating trackpad gestures, audio, and keyboard behavior.

lxj5820/dsh-boot-animation: Replaces the DSH boot screen with a full-window video that fades into the app.

The bitter lesson for computer vision: Vincent Sitzmann discusses how computer vision could become part of end-to-end perception and action loops.

Hadamard MLP: A channel-mixing design based on the Hadamard transform explores ways to reduce feed-forward layer parameters.

Qwen-Image-2.1: The official introduction page for Qwen-Image-2.1.

Tokenizers v1 benchmarks: Benchmarks for the Tokenizers v1 release candidate cover encoding, decoding, and parallel performance, with instructions for reproduction.

Grok 4.7: The official announcement describes Grok 4.7 updates for coding and knowledge work.

KDA Doesn’t Care About the Future: An investigation of future-dependent rounding in KDA kernels and whether models exploit that numerical effect.

Halo: A distributed training framework for Hugging Face models with checkpointing and multi-turn reinforcement learning.

Limite 1B — Violetto: Paradigma introduces a one-billion-parameter model for mathematical reasoning and describes its training and inference implementation.

MiMo-V2.6 · MiMo-V2.6 Technical Report: The official introduction to Xiaomi’s MiMo-V2.6 model series.

HySparse2: A hybrid sparse attention architecture with two-level key-value cache sharing for long-context and multi-turn agent tasks.

Jesse: An open-beta service for solidSF’s inference engine with an OpenAI-compatible API.

Batch Size Scaling — Tencent Hunyuan RL Team: The Tencent Hunyuan team discusses how batch size relates to training outcomes when scaling reinforcement learning for language models.

Introducing Claude Opus 5.5 | Anthropic: Anthropic introduces updates to Opus 5.5 in coding, knowledge work, communication, and operating cost.

Introducing GPT-6 Sol and Luna | OpenAI: OpenAI introduces GPT-6 Sol and Luna, covering improvements to capabilities, inference efficiency, and caching.

Qwen Intelligence: The Qwen Intelligence project website.

The Generalization Problem | Q: Explores combining model search with generalization priors so that learning can scale with compute.

Introducing HLE-Diamond: A cleaned and refined set of 1,000 questions from Humanity’s Last Exam, covering reasoning and expert knowledge.

Computational depth is all you need | Q: Studies the relationship between computational depth, model expressiveness, and training efficiency, with experiments on deeper language models.

Visualizing modular-addition training | Sauers · The models just want to learn | Sauers: Compares output distributions after 2,000 and 20,000 training steps, using circular plots to visualize learning in a modular-addition transformer. A second post shares the author’s thoughts on model learning.

Intelligence Density | Trajectory: Discusses intelligence density as task performance per unit of inference compute.

What my deep model doesn’t know | Yarin Gal · Dropout as a Bayesian Approximation: Yarin Gal introduces uncertainty estimation with dropout, alongside the paper interpreting dropout as approximate Bayesian inference.

Mathematics / Research / Learning

The Khachiyan constant and GPT-6 Astra: The author reports on using GPT-6 Astra to explore a conjecture about the Khachiyan constant.

Hilbert space: treating functions as vectors: An introduction to Hilbert spaces starting from the idea of functions as infinite-dimensional vectors.

Memristive singular value decomposition: The authors report singular value decomposition on memristive compute-in-memory chips and discuss applications including image reconstruction and model-weight decomposition.

Hamiltonian Neural Network with PyTorch: A tutorial on training Hamiltonian neural networks in PyTorch through an oscillator example.

Chinese Character Map, second edition · Chinese Character Map: Matrix67’s introduction to the second edition of the Chinese Character Map.

Optimization Constants in Mathematics: A collection of mathematical optimization constants and known upper and lower bounds, with links to research sources.

CIS 6270: Discrete Generative Models: A University of Pennsylvania course covering discrete generative models, diffusion, flow matching, and optimal transport.

MS&E 319: Efficient Generative Language Models: A Stanford course on training objectives, architectures, and inference algorithms under limited computational budgets.

Physics in the age of LLMs: Oz Amram shares his experience and views on how large language models are changing his physics research workflow.

Advisory Group on Mathematics and AI · AGMAI: OpenAI introduces an independent mathematics advisory group and collaboration on reviewing results, communicating research, and developing research tools.

A Proof of the Most Informative Boolean Function Conjecture: The authors propose a computer-assisted proof of the Courtade–Kumar conjecture, with computational verification records and a description of formal verification.

Dictators are most informative: The authors propose a proof of the Courtade–Kumar conjecture concerning the information retained by dictator functions under binary noise.

Unsolved problems in mathematics: A Wikipedia index of open problems by mathematical field, with references for further reading.

devol.dev: Interactive tools for learning robot kinematics, control, motion planning, and design.

A Faster Shortest Path Algorithm | Vals AI: The author reports using Claude agents to develop the C-HD shortest-path algorithm and a Lean proof, improving an asymptotic bound for a specific graph-density range without establishing a practical speedup.

Quanta Magazine: A science magazine covering research in mathematics, physics, biology, and computer science.

Claude discovers a novel enzyme system | Anthropic · Research preprint (PDF): Anthropic reports that Claude identified an enzyme system with CRISPR-like repeats in sequence databases and describes experimental work; its function remains unknown.

Slava Krushkal’s good-group discovery | Logical Intelligence: Logical Intelligence reports on Slava Krushkal and the team’s AI-assisted study of a new good group in four-dimensional manifold topology.

mo271/Zeta5: A Lean 4 formalization project for the irrationality of $\zeta(5)$; the repository reports a completed proof and explains its differences from the preprint.

Interactive Linear Algebra — UBC edition: An interactive linear algebra textbook covering vectors, matrices, linear transformations, eigenvalues, and related topics.

Headlines and inside stories | Tapio Schneider: In a guest post on Terence Tao’s blog, Tapio Schneider discusses turning AI-generated results into understandable, auditable, and trustworthy mathematical and scientific knowledge.

We’re gonna need a lot more mathematicians | Amit Sahai: In a guest post on Terence Tao’s blog, Amit Sahai argues for expanding the mathematical research community to understand AI-generated ideas and preserve human agency.

Claude computes a nine-loop amplitude | Anthropic: Anthropic describes Claude’s calculation of a nine-loop six-particle amplitude in planar $N = 4$ super-Yang–Mills theory and how the result was checked.

A question: does the sequence $n + \tan(n),\ n \in \mathbb{N}$ have a lower bound?

Programming / Systems / AI Tools

GPUI Markdown Studio: A WebAssembly-based Markdown editor and preview with Mermaid diagram support.

LayerFS: LayerFS gives agents isolated disposable filesystem branches with checkpoints, rewinding, and state reuse.

Cloudflare Quick Tunnels: Run cloudflared tunnel --url http://localhost:8000 to create a temporary public URL for a local service and share previews or run tests.

Eli Bendersky’s website: Eli Bendersky’s technical blog covers programming languages, systems, and mathematics.

Reader: An open-source macOS reading workspace for PDF, EPUB, and TXT, with annotations, notes, and AI chat.

Tylina: A Typst editor for editing the typeset page directly, available on the web and desktop with AI agent integration.

The senior engineer death spiral: An essay on overwork driven by the urge to prove oneself, and on maintaining communication and delivery as a senior engineer.

CUDA Matmul Optimization Worklog: A step-by-step CUDA matrix multiplication worklog covering coalesced memory access, shared memory, and tiling.

Hallucination Translation: A Mac document translation app with side-by-side reading that preserves formulas, citations, and layout.

OpenCode Reloaded: An explanation of how OpenCode hot-reloads configuration, tools, and plugins while sessions keep running.

Step Code: A terminal coding agent with code editing, tests, plugins, multi-agent coordination, and static-site publishing.

Nowdex: Tracks remaining usage for several AI tools across Mac, iPhone, and iPad, with a menu bar app, widgets, and iCloud sync.

mujocolab/mjlab: Provides an Isaac Lab-style interface powered by MuJoCo-Warp for reinforcement learning and robotics research.

iAmCorey/Wake: An agent session manager built with Rust and GPUI for browsing, searching, and resuming sessions.

heypcb · Circuit World: Generates schematics, board layouts, and fabrication files from natural-language descriptions, with KiCad projects you can keep. Circuit World showcases user-published boards and open-source hardware.

ReShiki: A tool for drawing molecules, building reaction schemes, and preparing chemical figures.

magpie: Selects models for agents such as Codex and Claude Code through a menu bar app and a local gateway.

Synara: An introduction page for an AI coding workspace for Claude Code, Codex, and Cursor.

Empryo: Maintains a live map of a code repository and reads it before edits, with desktop and terminal interfaces.

o8vm/octox: A Unix-like operating system written in Rust and inspired by xv6-riscv.

AGIdock: A cloud platform for agents that provides infrastructure for running applications.

yu1745/wetype-ime-linux: Ports WeType’s Android input engine to Linux and integrates it with Fcitx 5.

Lorca: Runs AI teammates on your computer to work on files and tasks through individual or group chats.

Using Claude Code: Spending your effort | Thariq: Uses evaluations and practical tasks to examine how reasoning effort affects speed, independent judgment, and verification of edge cases.

kaolti/phantomat: An infinite-canvas window manager for Hyprland with zoomable window layouts.

Design / macOS / Other Finds

Working Title: Working Title is an app for Mac.

Crisp: An open-source macOS display utility for scaling, brightness, volume, and display presets.

21st: A collection of React components, page templates, and themes with previews and implementation code.

HowToLiveBetter: A collection of health, everyday-life, and public-service resources with costs, benefits, and references.

Powermove: A video editor that connects to coding agents and supports layers, keyframes, and custom extensions.

Inspora: A curated archive of recent visual design and creative work.

shawn.: Design engineer Shawn’s portfolio features clean interfaces and interaction details.

Opensource UI: A copy-and-paste React and Next.js component library with interface controls and device mockups.

quicker.chat: A Mac inbox that brings direct messages together and offers conversation organization and reply drafts.

MiMo Gallery: A MiMo showcase presented as a three-dimensional portfolio museum.

Top-Conf Figure Gallery: A gallery of paper teaser figures that can be browsed by conference, year, and visual pattern.

Pelican on a Bike: A real-time 3D demo of a pelican cycling along the coast, with waves, a day–night cycle, and music.

Variora: Compares interactive creations from different AI models given the same brief.

Designeer: A curated collection of interface design tools, component libraries, design systems, and inspiration.

TinyWeb: A small native Mac browser with a borderless interface, vertical tabs, and searchable history.

Search | Office Commun · driceroland/Search: A lightweight macOS browser built on WebKit, with a minimal interface and a focus on quick startup.

Lean Browser: A minimal macOS browser built with Swift and WebKit, offering Zen mode and stating that it collects no telemetry.

Macfolio: Curated software, hardware, workspace setups, and reading for the Mac ecosystem.

Sidekick: Adds a second dock to the Mac for apps, links, clipboard history, and live widgets.

saragordic/rooms: Saves each project’s window layout as a room and switches between workspaces with a shortcut.

PriceRadar: Compares download, in-app purchase, and subscription prices across App Store and Google Play regions, with price alerts.

Reward is not the optimization target — LessWrong: Explores the distinction between reinforcement learning rewards and the goals a model learns. In an update, the author acknowledges that reward can empirically become a primary goal, though this does not follow by definition.

AI Alignment Forum: A community for AI alignment research and discussion.

Bayes’ rule — LessWrong: An introduction to Bayes’ rule and the intuition behind it.

LessWrong: Articles and discussions about rationality, cognition, and AI.

Dario Amodei — We Must Pace the Frontier: Dario Amodei discusses coordinating the pace of frontier AI development to give safety research and society more time to respond.

GPT 6 Astra Embodied Policy Evaluation: An anonymous evaluation report on GPT-6 Astra’s embodied policies.

Latest research - Goodfire: Goodfire’s research directory, focused on model interpretability and internal mechanisms.

Augmented Lagrangian Predictive Coding: training 1000-layer networks without backpropagation: Sakana AI applies an augmented Lagrangian method to predictive coding, exploring ways to train deep networks without backpropagation.

P(doom) | Armin Ronacher’s Thoughts and Writings: Armin Ronacher reflects on catastrophic AI risk and how people express its probability.

The Age of Wonders and Terrors | Scott Aaronson: Scott Aaronson discusses the hopes and concerns raised by AI progress.

Introducing System One Models & Jev | TypeSafe AI: TypeSafe AI introduces System One models and Jev for structured decision-making.

From Academia to Alignment | Mike Winer: Mike Winer shares his experience and reflections on moving from academic research to AI alignment.

Existential Risk from AI: An Exposition for Mathematicians: An explanation of AI existential risk and alignment for readers with a mathematical background.

MiMo-v2.6 RL: Xiaomi’s introduction to reinforcement learning for MiMo-v2.6.

Introducing Arrow 2 and Arrow 2 Telos – QuiverAI: QuiverAI introduces the Arrow 2 and Arrow 2 Telos vector graphics generation models.

Training Search Agents with GRPO — Jasper Lu: A practical account of training search agents with GRPO.

Helix 2.5: Zero-Shot 30-Home Generalization: Figure evaluates Helix 2.5 on three household tasks in 30 unseen homes, reporting an overall success rate of 56%; the tasks were fine-tuned in other environments.

Cache-to-Cache: Direct Semantic Communication Between Large Language Models: Research on transferring intermediate semantic representations directly between language models through KV caches, reducing the information loss of text-only communication.

Pretraining data, not verifiability, is why LLMs are especially good at math (and coding) — LessWrong: Proposes that the structure of pretraining data may explain language models’ strengths in mathematics and programming better than the verifiability of answers.

Reading a Pretraining Run — Chunyuan Deng: A guide to monitoring loss, gradients, activations, and evaluations during pretraining, including metric definitions, sampling methods, and diagnostic priorities.

Qwen3.8-livetranslate: The official introduction to Qwen3.8-livetranslate.

Mathematics / Research / Learning

Stanford CS 312 | Deep Learning Alchemy: A Stanford course on practical deep learning training and experimental methods.

Scaling law tutorial: A tutorial on neural network scaling laws.

A beginning for mathematics · Daniel Litt: Daniel Litt discusses the future of mathematical research in the age of AI and the value of understanding, communication, and new questions.

Professor Bingsheng He: Professor Bingsheng He’s academic homepage.

AI in Science | Google: Google’s report on AI’s role in advancing scientific research.

Building Labs that Learn – Periodic Labs: Periodic Labs outlines a research approach in which laboratories keep learning from experimental feedback.

Nature Is Our Learning Environment – Periodic Labs: Discusses using feedback from real-world experiments to train models.

AI Infrastructure at Periodic – Periodic Labs: Introduces the AI infrastructure supporting scientific agents, experiments, and training.

Fast math/slow math | Benjamin Antieau: Benjamin Antieau discusses two rhythms of mathematical research: solving problems quickly and building understanding slowly.

Conjectures.io: Expresses open mathematical conjectures as precise Lean statements and accepts verifiable proofs or counterexamples.

A Short Walk Through the Garden of Mathematics: Matrix67’s popular mathematics book, A Short Walk Through the Garden of Mathematics.

The Fate of the Riemann Hypothesis: Revisited – Proofs and Prompts: Richard Evan Schwartz revisits his short story about the Riemann hypothesis and discusses AI’s effects on mathematical research, education, and careers.

Top 500 Open Problems by LLM-assessed Importance — ProofAtlas: Ranks 500 open problems by importance as assessed by language models; the ranking does not represent a consensus among mathematicians.

Sur la destruction des maths par les boîtes d’IA | David Madore: David Madore comments on the relationship between AI companies and mathematical research.

Solving a Rubik’s Cube with graph theory | The Math Flow: A post explaining Rubik’s Cube solving through graph theory.

An Introduction to Universality | Alessandro Codello: Alessandro Codello’s introductory lectures on universality, covering fractals, percolation, chaos, and phase transitions.

Impossible Questions | Scientific American: A Scientific American collection exploring the origin of life, longevity, dark matter, consciousness, and the beginning of the universe.

I compiled these articles into Chinese–English bilingual PDFs, with separate layouts for phones and computers: Mobile edition (PDF, 103 pages) · Desktop edition (PDF, 56 pages).

• Finding meaning in the impossible (editor’s introduction): Pursuing seemingly unanswerable questions can lead to unexpected practical discoveries; the journey itself may be meaningful.
• The questions science may never answer: Science offers approximations that keep evolving. Some questions may remain out of reach because their subjects cannot be observed, or because observation changes them.
• Will we ever know how life began?: Experiments can reproduce possible paths to life’s origins, such as metabolism-first scenarios or an RNA world, but establishing which path actually occurred is much harder.
• Will we ever achieve immortality?: Explores slowing aging and cellular reprogramming. Current longevity records do not establish an unbreakable biological limit, though substantially extending life remains a major challenge.
• What if we never find dark matter?: Gravitational evidence for dark matter is extensive, but its particle identity has not been directly confirmed. Perhaps science can offer predictions without reaching an absolute truth.
• The hunt for reality’s smallest pieces: Could strings or quantum units of spacetime lie beneath the Standard Model’s 17 particles? Experiments approaching the Planck scale remain almost out of reach.
• What if ‘consciousness’ isn’t real?: Consciousness may not be a useful scientific category; studying infants, bees, or AI might call for different distinctions.
• What came before the big bang?: No-boundary models, a big bounce, and a CPT-mirrored universe offer testable predictions, but “before” may be meaningless or impossible to access directly.
• Why we love unexplainable mysteries: An interview with Unexplainable about the appeal of mysteries, from the universe’s origins to why bicycles stay upright. The unknown need not be discouraging.

Programming / Systems / AI Tools

lulu-loopp/folio-terminal: A terminal for Windows and macOS that brings formula rendering, file previews, and agents’ waiting status into the workspace.

zyycn/codex-proxy-rs: A self-hosted transparent proxy gateway for multiple Codex accounts, written in Rust.

My AGENTS.md: A Life Spent Wrestling with Unnatural Prose and Overengineering: Notes on using AGENTS.md to guide models’ writing and curb overengineering.

Why TeX is Slow and How We Rebuilt It in Pure Rust | Leo Liu · leoliu0/ratex: An introduction to TeX’s performance bottlenecks and RaTeX, a typesetting engine rewritten in Rust.

MiniMax-AI/minimax-code: An open-source terminal coding agent powered by MiniMax.

Ryoku-dev/ryoku-arch: An Arch Linux desktop distribution combining Hyprland, an installer, and system configuration.

The last mile of a long road: faster NumPy in the browser | Notebook.link: The engineering work behind making NumPy run faster in the browser.

UXSpice: A circuit simulator built with Svelte, WebGPU, and ngspice compiled to WebAssembly.

Design / macOS / Other Finds

Qingjian Input Method: Adds foreign-language vocabulary to Chinese typing, helping users pick up words along the way.

Design Studio AI: A design workspace built for agents.

The Component Gallery: Organizes examples from real design systems by component type, making it easier to compare implementations.

Libraries.dev: UI effects libraries that coding agents can install and configure, including glowing borders and liquid effects.

saragordic/window-sweaters: A macOS menu bar app that adds knitted borders to windows.

livabble: A guide to cities and everyday life for Chinese-speaking digital nomads.

KiwiCanopy/KiwiDesk: A macOS tiling window manager with multiple layouts configured through a settings interface, while preserving native desktops.

Savant: The Savant project homepage.

jaywcjlove/key-clicker: Adds mechanical keyboard and typewriter sounds to typing on macOS.

ObsidianUI: A component library built with React and Tailwind CSS.

Compositor: An image editor for Mac.

Design inspiration bookmarks:

• Mobbin: A large collection of real app interfaces for studying design patterns.
• Godly: A web design inspiration gallery, now redirecting to Recent.
• Awwwards: Website awards recognizing outstanding design and execution.
• Cosmos: Collect and organize visual inspiration, with search by color and visual similarity.
• Curated Design: A curated collection of websites organized by category.
• Design Spells: An archive of delightful interface details.
• 60fps: A gallery focused on interaction and polished animation.
• Supahero: Website hero sections for quick inspiration.
• Saaspo: Design inspiration for SaaS websites and landing pages.
• Layers: A showcase of designers’ work.
• Minimal Gallery: A handpicked web design gallery.
• MNMM: An archive of minimalist design.
• Search System: An archive of graphic and editorial design.
• Rebrand Gallery: Rebranding projects with before-and-after comparisons.
• Seesaw: A curated web design gallery updated daily.
• Same Energy: Image search based on visual similarity.
• Spiral: A spiral visual feed for discovering inspiration.
• Game UI Database: A database of game interfaces organized by genre.
• Interface in Game: Game interfaces organized by element type.
• HUDs and GUIs: Futuristic interfaces from films and games.
• Pinterest: A longstanding favorite that remains useful.
• Loadmo: An archive of creative mobile web design.

Astra vs Fable 5.1: Surya shares outputs from the two models given the same prompt.

Grok Bot Marketplace: A directory of bots and skills for Grok Bot.

Maze Bench — AI Leaderboard: Compares AI agents exploring mazes, with rooms visited, gems collected, token usage, and run replays.

Introducing ChatGPT Images 2.5 | OpenAI: OpenAI introduces an updated image generation and editing model, highlighting improvements in lighting, textures, reference fidelity, and editing across multiple turns.

GPT-6 Astra, Looped Transformers, and Hidden Reasoning: Sebastian Raschka discusses looped Transformers and hidden reasoning in the context of Astra; the architectural speculation remains unconfirmed by OpenAI.

DeepSeek V4.1 Flash: Stronger, Faster, More Accessible: DeepSeek introduces V4.1 Flash with updates to model capabilities and inference efficiency—let’s go!

Introducing SWE-2: Pushing the Pareto Frontier | Cognition: Cognition introduces SWE-2 for software engineering, focusing on the trade-offs between coding ability, speed, and cost.

>10x More Efficient Pretraining — Magic: Magic reports more than a tenfold improvement in pretraining efficiency when comparing held-out loss against training compute; this is the team behind 100M Token Context Windows — Magic.

Pacing the Frontier: A public statement by employees of frontier AI companies calling for coordination on the pace of AI development to allow time for safety work and societal preparation.

AI researchers debate how close we are to recursive self-improvement: A discussion among researchers about recursive self-improvement, progress in automating AI research, and the remaining bottlenecks.

Mathematics / Research / Learning

Robot Learning | ETH Zürich: A course covering imitation learning, reinforcement learning, and policy optimization through to vision-language-action and robotics foundation models, with slides, recordings, and practical assignments.

Moon | Bartosz Ciechanowski: Interactive explanations of the Moon’s orbit, tidal locking, phases, and solar and lunar eclipses.

Tristan Buckmaster’s statement: Buckmaster describes results on fluid-equation blowup under smooth forcing obtained with Levent Alpöge using LLMs, alongside the research process and publication dispute; subsequently, On the Navier–Stokes Millennium Prize Problem | OpenAI announced a proof and Lean formalization of finite-time blowup for the three-dimensional Navier–Stokes equations under smooth forcing, stating that it establishes formulations C and D of the Millennium Prize Problem.

AlphaGenome | Google DeepMind: Predicts molecular properties related to gene regulation from long DNA sequences and assesses variant effects by comparing predictions before and after a mutation.

The geometry of nonergodic composition — Belief Updates: A discussion of the geometry of nonergodic composition—quite elaborate, and I haven’t made sense of it yet.

Declaration — Math and AI: A declaration on AI and mathematics that emphasizes conceptual understanding, sharing methods, and training researchers, while questioning progress measured by the number and speed of solved problems.

Programming / Systems / AI Tools

datawhalechina/zero-to-sglang: A bilingual inference course from SGLang and Datawhale, progressing from inference fundamentals and building mini-sglang to reading the production codebase.

FeiZhuLulu/real-api-pricing: Compares API costs by dividing subscription fees by usable tokens, with allowances, unit prices, and Pareto charts.

CheshireMew/PuppetLoom: Turns layered character PSDs into animated 2D characters with rigging, validation, and calibration, allowing external agents to participate through a CLI.

kevinzakka/mjbatch: A Python library for running large batches of MuJoCo simulations in parallel on CPUs.

Games | OpenAI Developers: A showcase of games built with OpenAI tools, with projects and development write-ups.

Rapidly scaling online storage to serve over 1 billion ChatGPT users | OpenAI: OpenAI shares engineering practices for scaling ChatGPT’s online storage.

Design / Productivity / Other

realchendahuang/feedsieve: A Chrome extension for filtering spam accounts on X, highlighting high-confidence matches and supporting native blocking with one click.

The Death Of The Button | Smashing Magazine: Explores how AI and context-aware interactions can reduce explicit interface actions, along with the design trade-offs involved.

Countering misuse of AI: September 2026 | Anthropic: Anthropic presents its investigation into AI misuse, describing cases involving cyberattacks, fraud, and unauthorized distillation, along with its response.

Design Words: Uses visual examples to help select and combine design vocabulary for more specific interface prompts.

Quantization-Aware Healing: a compressed, 4-bit model that outperforms its full-precision original: Multiverse Computing introduces a 4-bit quantization-aware healing method that outperforms a recovered 60B BF16 checkpoint of the same architecture on seven of its nine reported benchmarks. The additional distillation training was not matched, so the result cannot simply be attributed to quantization itself.

Introducing the XGBoost Vector-Leaf Model | XGBoost: An introduction to trees whose leaves output vectors, using shared tree structures for multi-target prediction.

[2608.27370] Puro-2B: Poor Lab’s Qwen2-1.5B Trained on RTX 5090 within $5090: An exploration of low-cost small language model training on RTX 5090 GPUs. The best model in v2 cost less than $6,900 to train; the estimate of reaching Qwen2-1.5B performance for about $4,400 comes from fitting scaling laws.

Training frontier knowledge work agents: A 397B RL training guide with SkyRL | Mercor Blog: Mercor shares engineering experience from training a 397B knowledge work agent with SkyRL.

chenglou/sotaku, Sotaku v2 solves 99.12% of a 25,000-puzzle Sudoku benchmark with an 800K-parameter looped transformer trained from scratch: A roughly 800K-parameter looped Transformer trained from scratch to solve Sudoku. The repository reports that Sotaku v2 solves 99.12% of a 25,000-puzzle benchmark using FP32 and 1,024 inference iterations.

Z1T | Transformer-like models for Z1: Sparse Transformer-like models designed for Z1 probabilistic hardware, exploring model and hardware co-design while sharing some training recipes and weights.

GPT-6 Astra on robot arms | Robocurve: Experiments with GPT-6 Astra operating robot arms. Watching this makes me a little more inclined to believe LLMs can solve everything, though what a demo can actually do still comes down to the specific task.

Model guidance | OpenAI API on using Astra.

**Using the model**
- Use the Responses API and set `model: "gpt-6-astra"`
- New capabilities: asynchronous tool calling (set `async: true` on the tool definition), changing instructions mid-turn (WebSocket mid-turn steering), and changing reasoning effort with `configuration_update` in standard single-agent requests (keeping the request-level effort preserves the original prompt prefix)
- `reasoning.effort: "none"` is unsupported; Fast mode is unavailable with EU data residency

**Migration notes**
- If you previously used `none` / `minimal`, try `low` first
- Prefer Responses for tool use; Chat Completions is available, but tool calling requires Responses
- Remove `temperature` / `top_p` / `top_logprobs` (and the corresponding logprobs)
- When migrating from GPT-5.5 or earlier, update caching: `prompt_cache_retention` → `prompt_cache_options.ttl: "30m"`

**Prompt tuning (the most practical part)**
1. **Initiative**: It tends to ask more follow-up questions by default. To have it finish the work autonomously, explicitly say “infer intent, bias toward action, and don't stop at acknowledging capability or delivering partial work.” Produce a reviewable result before asking for approval.
2. **Instruction priority**: It is more sensitive to skills / `AGENTS.md`; state that “user instructions take precedence over skills.” If it gets stuck, have it quote the specific skill rule.
3. **Writing style**: It tends toward lists, tables, and stock phrases by default. If you want prose, specify paragraphs first, fewer nested lists, and no slop (delve, Bottom Line, X not Y, etc.).
4. **Subagents**: It may delegate too little by default; specify when to delegate in parallel for multi-agent work.
5. **Testing**: It tends to overtest code. For small changes, state “don't write tests that mirror the implementation for reversible, low-impact changes.”

What I focus on in practice: make autonomous execution, writing style, parallel delegation, and testing boundaries explicit to reduce unnecessary questions.

eric provencher on X: “Rethinking skills and prompts for GPT-6 Astra” / X: Eric Provencher discusses writing skills and prompts for Astra.

Mathematics / Research / Learning

The Math of RL - From Policy Gradients to PPO and GRPO: An article on the mathematics of reinforcement learning, from policy gradients to PPO and GRPO.

Introduction to Robotics | IRoM-Lab: Princeton’s introductory robotics course, covering planning, control, state estimation, and robot learning, with drone project assignments.

[Blog] Estimate Diagonal Hessian: An article on estimating diagonal Hessian entries.

Axiom: An AI project for mathematical reasoning and formal proofs.

Announcing FrontierMath Erdős | Epoch AI: Epoch AI introduces a mathematics benchmark of 68 open Erdős problems, using Lean to verify proofs or disproofs.

francesco215.github.io/autoregressive_diffusion/uncertainty.html: Reading on autoregressive diffusion and uncertainty.

\[2609.02867v2\] Magnetized accretion onto rapidly spinning binary black holes: mini-disk thermodynamics, magnetic transport, and dual jets: Numerical simulations of magnetized accretion onto rapidly spinning binary black holes, including mini-disk thermodynamics, magnetic transport, and dual jets.

How Our Data Shaped Neural Architecture Discovery, and How Automation Can Reshape the Future | Core Automation: The article discusses data and the inductive biases of neural network architectures. A randomly initialized, frozen network may still provide useful features, though the linear classifier in the figure was trained.

This makes me wonder: could feature quality across multiple random initializations help screen architectures for a particular type of data in advance? We need to distinguish the quality of the random features themselves from the contribution of subsequent classifier training.

A neural network architecture with a suitable inductive bias may map inputs into a feature space that a simple classifier can use more easily, even when its main weights are random and frozen. Training can further align this initial structure with a specific task.

Figure. Output probabilities of a (linear) logistic classifier trained on a 10,240-dimensional NRF(Neural Random Feature) representation, extracted from identical, randomly initialized ResNet-18 architectures.

Some related papers:

• [2410.04368] Algorithmic Capabilities of Random Transformers: Freezes the random Transformer body while still optimizing embeddings to activate algorithmic capabilities.
• [1711.10925] Deep Image Prior: Uses network structure as an image prior without a large training dataset, but still optimizes for an individual image.
• [1906.04358] Weight Agnostic Neural Networks: Searches for network architectures that perform well across multiple shared weight values.
• [2206.08464] PRANC: Pseudo RAndom Networks for Compacting deep models: Freezes multiple random basis networks and learns their linear combination coefficients to compress a model.

Theory of Mind as a Central Skill for Researchers: Treats theory of mind as a meta-skill that spans writing, presenting, reviewing, and collaboration. First consider what the audience knows, cares about, and might misinterpret, then decide how much to explain and in what order.

• When presenting, pair abstract concepts with examples, repeat important points, and avoid large blocks of text. Consider the occasion and the audience.
• When writing a paper, state your points clearly, give figures and tables detailed explanations, make the whole paper easy to browse, and minimize abbreviations. Avoid adding to the reader’s memory burden.

Arena Physica Thesis: Electromagnetic engineering faces bottlenecks in solvers, fragmented CAE, EDA, testing, and field data, as well as slow and expensive validation feedback. Arena Physica proposes connecting design, simulation, and testing through agent tools, physics foundation models, and real laboratory data.

[2608.29530] The Emergent Symbolic Structure of Artificial Neural Networks: Trained neural networks may implicitly contain symbolic structures, suggesting that symbolism and connectionism may not be entirely opposed. Related paper: [2506.21797] Why Neural Network Can Discover Symbolic Structures with Gradient-based Training: An Algebraic and Geometric Foundation for Neurosymbolic Reasoning

Atlas Fields Studio: Arena Physica’s interactive electromagnetic field prediction demo lets you modify geometry and observe predicted changes in electric and magnetic fields. It is currently labeled Beta.

What is a Smith Chart? | Arena Physica: An introduction to Smith charts in RF engineering.

Programming / Systems / AI Tools

Your First GPUI App - Building a Desktop UI in Rust | Mrsheerluck Blog: An introductory tutorial on building your first desktop application with Rust and GPUI.

emilkowalski/skills: Skills for Designers and Engineers.: Emil Kowalski turns his interface design and animation experience into agent skills covering animation implementation, review, interaction details, and UI prototypes.

DEEIX-AI/DEEIX-Chat: An enterprise AI workspace for model routing, multimodal chat, files, tools, billing, identity, and operations.: An enterprise AI workspace integrating model routing, multimodal conversations, files, tools, billing, and identity management.

Zodex | AI integrated development environment: An AI development workspace built with Rust and Tauri that brings agent conversations, terminals, Git diffs, and rollbacks together. It can import local Codex, Claude Code, and Pi sessions.

Skillry - Agent Skills with Taste: A curated agent skills catalog covering websites, slides, images, and video, with examples and installation instructions.

Polars — Pre-release of Polars 2.0: An introduction to the first release candidate of Polars 2.0, which makes the streaming engine the default for LazyFrame queries and tightens implicit behavior such as type coercion.

get-bb/bb: The agent IDE that builds itself: An agent IDE that can be driven through a desktop app, web interface, CLI, or HTTP API, with live task tracking and steering.

xuanji86/ainiee-translate-skill: An independent, agent-native novel translation system that works similarly to AiNiee without depending on its application, using Claude Code as the translation engine.: A novel translation skill powered by Claude Code that runs independently of the AiNiee application.

mishka-zz/freecad-microwave: FreeCAD workbench for RF and microwave engineering: A FreeCAD workbench for RF and microwave engineering, still in developer preview. Its currently available simulation backend is openEMS.

Blueprint - AI Hardware Design Tool: An AI-assisted hardware design tool.

KiCad - Schematic Capture & PCB Design Software: Open-source software for schematic capture and PCB design.

American-Embedded/kistack: KiStack is a HUMAN WRITTEN bunch of skills for KiCad: Agent skills for KiCad workflows.

TrenTorch: An educational project inspired by Harvard TinyTorch that teaches PyTorch by building a machine learning framework from scratch.

Design Engineer Tools: A tool directory for design engineers.

DesEngs: Resources for Design Engineers · Minimal Website Design Directory — DesEngs: Directories of design engineering resources and minimal website references.

on.design: A resource project for designers, currently accepting sign-ups ahead of launch.

whatships.com — Discover startup launch videos · Tools for making launch videos or mockups— whatships.com: A collection of startup launch videos and tools for making videos and mockups.

Icon and design resource bookmarks:

• Remix Icon - Open source icon library
• Phosphor Icons
• Tabler Icons
• Blode Icons (GitHub) · Blode Icons
• Iconify · Iconify icon sets
• Untitled UI
• Heroicons
• Keyline Icons
• IconMonstr
• Bootstrap Icons
• Reicon
• Lucide
• Icônes
• icons0
• Mage Icons
• MingCute
• Hugeicons
• Gravity UI
• Central Icons
• Solar Bold Icons
• SVG Repo CC Attribution
• Streamline Pixel Icons
• toools.design
• Silk icons (Wikimedia)
• @huge_icons
• @nucleoicons
• @antfu7

macOS / Productivity / Other

jurplel/InstantSpaceSwitcher: Native space switching on macOS with no animation: A tool for disabling macOS’s native Space switching animation.

Omakade | Your games, beautifully together: A local game library for Omarchy that brings together installed games from Steam, Lutris, Heroic, and other sources, then hands launching off to their original platforms.

Download Codenotch — Vinz: A small tool that displays LLM usage beside your Mac’s notch.

thisisgm/flea: A fast, keyboard-first file manager for Omarchy: Quickshell front end, Rust backend.: A keyboard-first file manager for Omarchy with a Quickshell frontend and Rust backend.

omafiles — the file explorer Omarchy was missing: An Omarchy file manager built with Rust and GPUI, integrating search, previews, and Git diffs.

Droppy. Your Mac, just way better.: Brings a file shelf, clipboard, media controls, widgets, and more to your Mac’s notch or a floating Dynamic Island.

Tenacity: An open-source multitrack audio editor and recorder with support for effects plugins, scripting, and signal analysis.

ChatGPT Training: OpenAI’s step-by-step exercises and workflow tutorials for completing practical tasks with ChatGPT Work and Codex.

Which AI Made This?: Compares different models’ responses to the same UI design task and shows the difference made by enabling design skills.

Shortap — Quick tap gestures for your MacBook: A gesture tool that triggers actions when you tap your MacBook’s chassis.

Related open-source projects:

• Tapify: A menu bar alternative to Knock, supporting single through triple taps for media controls, screenshots, and launching apps
• Bonk: GPL-licensed, supports up to four consecutive knocks, can trigger Shortcuts / shell commands, and allows different mappings for each app
• nocnoc: Includes a waveform display and calibration wizard
• knocker: Distinguishes left, center, and right taps and provides a menu bar interface
• TapCraft: A Python tool for chassis taps and trackpad zones; accelerometer detection requires sudo, while microphone detection does not

How Speculative Decoding Makes LLMs Go Brrr: Explains speculative decoding through its draft-and-parallel-verification loop and compares approaches including independent draft models, Medusa, MTP, EAGLE, DFlash, and DSpark.

Training AI to Paint with Code: Surya Narreddi’s team used reinforcement learning to train a language model to create editable p5.brush JavaScript paintings, focusing on aesthetic rewards built from hand-rated references and pairwise judgments.

deepManifold: Brings together two preprints and related essays that use fixed points, stacked piecewise manifolds, and boundary-conditioned iteration to formulate the geometry, algebra, and computation of neural networks.

GPT-5.6 Sol Runs Doom in Turing Complete: According to the creator’s demo and Tom’s Hardware, GPT-5.6 Sol built a virtual RV32IM CPU called Codex-R32 inside Turing Complete and ran a PureDOOM build on it, although the result was not yet playable.

First Results from Jalapeño: OpenAI reports that its first custom inference chip improved both power efficiency and end-to-end latency against the tested comparison systems while serving several large models in InferenceX; the chip remains in production validation.

GLM-5.3-Flash: Z.ai’s first natively multimodal GLM-5 model uses 320B total parameters, 18B active parameters, a hybrid linear-and-sparse attention architecture, and IndexPool to reduce indexing overhead at long context lengths.

Qwen3.8-Flash-Next: An open-weight multimodal MoE model and an early preview of the Qwen4 architecture, introducing hybrid GDN–QSA attention, gated residuals, and N-gram embeddings.

Hy4 Preview: Tencent Hunyuan’s open-weight flagship MoE preview has 770B total parameters, activates 49B per token, supports a 1M-token context window, and uses Gated DSA, IndexCache, and iHC.

Mathematics / Research / Learning

Proswit — Interactive Math Articles: A collection of interactive articles on probability, analysis, and foundational mathematics, covering topics such as Bayes’ theorem, the central limit theorem, Brownian motion, and Fourier series.

Could a Neuroscientist Understand a Microprocessor?: Jonas and Kording applied standard neuroscience analyses to a simulated MOS 6502 microprocessor and found that, although the methods produced patterns resembling neuroscience results, they still failed to recover the processor’s known hierarchical structure and functional mechanisms.

Symbolica: A high-performance computer algebra library for Python and Rust that manipulates large symbolic expressions, performs pattern matching, and generates optimized numerical code; its source is viewable, but redistribution and professional use are license-restricted.

s2 CLI and Skill: A read-only Semantic Scholar CLI and agent skill from Hugging Face for searching papers and retrieving references and citations, also used by the Papers with Code chat interface.

You Are NOT Dumb, You Just Lack the Prerequisites: A personal reflection arguing that difficulty with math and other complex subjects often comes from missing prerequisites, and that rebuilding the foundations can make progress possible.

Palomar: A public, searchable registry of Lean formalizations that pins source revisions, checks proofs with Lean, and publishes the exact statements, dependencies, and review comments.

Programming / Systems / AI Tools

GPUIX: Provides React and TypeScript bindings for Zed’s GPUI, rendering native desktop interfaces directly through Metal, DirectX, or Vulkan without Electron or web views.

Scrollbar but Cooler: A front-end experiment that redesigns the web scrollbar while preserving its positional cue and drag interaction.

Lody: A shared workspace for coding-agent teams that connects existing agents and machines through ACP and lets people share sessions and dispatch work from desktop, mobile, web, or CLI.

Archify: A Node.js rendering and validation system for coding agents that deterministically compiles typed JSON IR into self-contained HTML/SVG maps for architecture, workflow, sequence, data flow, and lifecycle diagrams.

Gribouille: A declarative Grammar of Graphics library for Typst, inspired by R’s ggplot2 and Python’s plotnine.

Markdown Graphs: A set of monospace chart and table components installable through shadcn and embeddable in React/MDX, with some components also offering ASCII versions for README files.

Fresh: A terminal text editor built around a zero-configuration, familiar interface, with mouse support, menus, a command palette, LSP, multiple cursors, an integrated terminal, and Git-worktree task orchestration.

Backend from First Principles: An open collection of backend-engineering notes and code examples spanning HTTP, routing, authentication, databases, caching, concurrency, observability, deployment, testing, and event streaming.

How Hugging Face Inference Endpoints, Jobs, and Buckets Power Search on Papers with Code: Hugging Face explains Papers with Code’s hybrid-search architecture: Jobs build paper embeddings offline, Buckets hand off artifacts, Inference Endpoints handle live queries and incremental updates, and RRF combines PostgreSQL full-text search with pgvector retrieval.

macOS / Productivity / Other

omacosy: An Omarchy-style desktop setup for macOS 26 on Apple Silicon, combining AeroSpace or OmniWM, Karabiner, and custom Swift helpers for dwindle tiling, a Super key, a status bar, focus-follows-mouse, and workspace gestures.

OmniWM: An open-source tiling window manager for macOS 26 and Apple Silicon, inspired by Niri and Hyprland and featuring Niri and Dwindle layouts, a workspace overview, a command palette, and optional IPC/CLI control.

Try Omarchy: Packages an ARM64 Arch Linux image configured with Omarchy Quattro, a QEMU runtime backed by Apple’s Hypervisor Framework, and a Swift/AppKit launcher into a macOS app for Apple Silicon.

AeroSpace Guide: The official end-to-end guide to AeroSpace, covering installation, TOML configuration, layouts, workspaces, keybindings, CLI commands, and multi-monitor behavior on macOS.

Workout Guide: An open workout illustration library with 302 exercises and 906 transparent SVG frames, searchable by equipment, muscle group, and exercise type, with MIT-licensed code and CC BY-SA 4.0 assets.

Blue Fish Archive: An unofficial fan-made archive of DeepSeek whale-girl stickers, with a browsable sticker wall, full-size previews, copying, and downloads.

math-ai-org/mathcode: A terminal-based mathematical programming agent with Lean integration that can inspect proof goals, search declarations, verify candidate proofs, and complete formal proofs interactively.

Cua: GUI sandboxes and scalable computer clusters for computer-use agents, with both local and hosted deployment options.

Introducing Delta: Zed’s multi-agent coding and review environment, synchronizing code, conversations, and worktrees in real time so people and agents share the same development context.

PyTorch: a reference language: A proposal to retain PyTorch as an executable reference specification, with LLMs generating explicit forward and backward implementations and verifying their equivalence. It also raises the question of whether LLM-written backpropagation could replace automatic differentiation in some settings.

wibus-wee/cradle-app: A desktop-first AI agent management platform for running agents, tracking tasks and sessions, coordinating workflows, and connecting multiple model providers in one place.

Building a distributed training framework from first principles: Umar Jamil explains how to build a distributed model-training framework from first principles; this description is based only on the public video title and author information.

Ornith-1.5: From Self-Scaffolding to Self-Improvement: A model that generates its own training tasks, task scaffolds, and solution trajectories, then uses reinforcement learning to form a continuous self-improvement loop.

Famous Deep Learning Papers: A chronological guide to classic deep-learning papers, summarizing the central problem and historical impact of each work.

CS 8803 - LLM: The course page for Georgia Tech’s Spring 2026 large language models class, collecting lectures, papers, and student presentations on pretraining, reasoning, alignment, and more.

Papers with Code: An index for browsing machine-learning papers, code implementations, and benchmark results by task.

CopilotKit/openbot: An open-source AI coworker platform that gives every agent an isolated browser, files, and tools while recording its decisions and actions.

Machine Studying: Introduces “Machine Studying,” in which agents autonomously develop domain expertise from a document corpus, and explores how to measure that ability with StudyBench.

Hot Take: LLM can ‘jump’: Argues that interconnected knowledge may let models make scientific leaps along paths that differ from human intuition, with implications for safety and continual learning.

[2608.08003] The Spectral Neuron: Introduces explicit nonlinearity through the eigenvalues of an affine matrix, seeking a balance among model expressiveness, coefficient transparency, and shape constraints.

Learn Inference: An interactive introduction to production inference engineering for generative AI, from attention kernels to multi-cloud capacity planning.

The Year in AI Papers: Essential Research from 2025–2026: A collection of important AI papers from 2025–2026, organized by topic, lab, citations, code, and date, with concise summaries.

Marin’s training launch: the Harrier K40 data overview shows the distribution of 23.11T tokens across 40 semantic domains and five quality buckets; the 535B-A23B 18T Token Hero Run report and public tracking issue document the model specification, scaling ladder, training dynamics, and risk-monitoring plan.

Mathematics / Research / Learning

Fanout: A technical learning platform collecting courses, roadmaps, paper explainers, and interactive experiments on AI research, system design, and the mathematics of machine learning.

Awesome Physics of AI: A learning-dynamics-oriented map of papers and research that use principles from physics to understand, predict, and explain deep learning.

[2608.16884v1] Improving the matrix multiplication exponent with modern optimization and AlphaEvolve: Combines reformulated optimization problems, machine-learning algorithms, and AlphaEvolve to improve the known upper bound for the matrix-multiplication exponent from 2.371339 to $\omega < 2.371177$.

Classic Patents: An open museum of historic technology patents with verified patent text, original PDFs, engineering explanations, and interactive 3D physical simulations.

How to Read a Technical Paper: Recommends a timed skim before close reading, while taking notes in your own words, questioning details, and considering extensions. LLMs can help summarize and retrieve later commentary, while careful reasoning should remain with the reader.

Lean formalization of bounded gaps between primes: A Lean 4 formalization of the bounded-gap theorem that infinitely many pairs of consecutive primes satisfy $p_{n+1} - p_n \leq 246$.

CS229 Course Notes: Stanford CS229’s machine-learning notes, systematically covering supervised learning, learning theory, unsupervised learning, reinforcement learning, and other foundations.

[2608.16753] Mathematics in the age of AI: Terence Tao discusses how the mathematical community should reconsider the goals and value of mathematical research as research-level AI arrives.

VibeMathed: A community collection of previously unsolved mathematical problems later proved or disproved with AI involvement, accompanied by verifiable sources and validation labels.

Platforms for formal mathematics: lemma.ing records mathematical problems, derivations, and research progress; Halmos checks user-submitted proofs with the Lean kernel; TheoremDB preserves open problems, proofs, failed attempts, and contributor attribution; ProofAtlas provides theorem explanations, Lean statements, and source code; and Proof.Fail maps the limits of LLM mathematical ability through failure cases.

Programming / Systems / AI Tools

cneuralnetwork/oplogs: A local-first experiment tracker for machine learning and agents that records metrics, artifacts, system telemetry, and LLM call traces without an account or cloud service.

Latias94/merman: A headless Mermaid parsing, layout, and rendering engine written in Rust, with output support for SVG, PNG, JPEG, PDF, and other formats.

AIcss: Copy-ready UI components for AI agent interfaces, covering thinking states, tool calls, streamed text, code diffs, and structured output.

lassejlv/gpui-starter: A minimal GPUI desktop application starter demonstrating Zed’s GPUI, theme configuration, native application entry points, and basic interaction.

huacnlee/omarchy-mihoro: Adds a Mihoro status-bar panel to Omarchy for monitoring proxies, switching operating modes, and managing multiple subscriptions.

Omarchy Plugins: An independent community plugin directory for Omarchy Quattro, with source links and copyable installation commands; the page contained no listed plugins when checked.

kenneth’s gitea: Kenneth’s self-hosted Gitea instance for publicly hosting and browsing Git repositories.

[2608.13759] GPU Offload in Rust: Portable, Safe, and Fast: Proposes a multi-vendor GPU compilation framework built into rustc and its LLVM backend, retaining Rust’s memory safety while approaching CUDA and HIP C++ kernel performance.

Getting Started with CPUs and GPUs in Rust: An open Japanese textbook for web developers that explains how CPUs and GPUs work through Rust code.

Linux Fundamentals: A free introduction to Linux and Unix covering the command line, file systems, processes, networking, Shell, and other fundamentals.

VibeHub: A plain-language, visual glossary of common Vibe Coding terms across frontend, backend, technology stacks, AI, and more.

Mojo is now fully open source · Mojo Manual: Mojo is now fully open source under Apache 2.0; the official manual covers its syntax, type system, memory model, and heterogeneous hardware programming, and is a useful starting point for comparing Mojo with Triton, CUDA, and TileLang.

OJO: A product workspace where multiple AI design agents collaborate to turn natural-language requirements into research, UI, interactive prototypes, code, and engineering handoff.

Git at any scale: Cursor reviews the storage and consistency challenges of large-scale Git hosting and introduces Continuity, an architecture centered on a write-ahead log in S3.

Computer History Preview: An experimental Cua Driver feature that stores encrypted action metadata and lets authorized agents query recent history through a read-only interface.

cathrynlavery/diagram-design: An editorial diagram skill for coding agents such as Claude Code and Codex, generating multiple kinds of self-contained HTML and SVG diagrams that can adopt a project’s brand style.

fx: A lightweight open-source coding-agent CLI written in Zig, emphasizing its roughly 6 MB size, fast startup, low context overhead, and embeddability.

Ladybird: An open-source browser project backed by a nonprofit and independently developing its browser engine and complete web platform stack.

Zellij.online: Shares Zellij terminal sessions directly on the web without SSH, port forwarding, or extra network configuration.

Briq: A native agent-assisted development workspace for Windows 11, designed to manage several coding agents in parallel.

OpenLogi: A local-first Logitech Options+ alternative written in Rust, using HID++ to remap buttons and control DPI and SmartShift without accounts or telemetry.

terminal-code · zenbu-labs/terminal-code: An open-source development tool that brings VS Code workspaces into the terminal.

debpalash/VoiceStudio: A fully local, open-source ElevenLabs alternative supporting voice cloning, dubbing, dictation, transcription, audiobook production, and more.

Subdomain Enumeration via Certificate Transparency: Passively enumerates known subdomains from public Certificate Transparency data and reports when each one first appeared.

fross100/omaplug: An independent Omarchy plugin manager for enabling, disabling, updating, installing, and removing plugins.

lijiawei0305-pixel/omarchy-mihomo-plugin: An Omarchy status-bar plugin for a standalone mihomo core.

Omarchy Manual in Chinese: A community Chinese translation of the official Omarchy manual, covering installation, configuration, keyboard shortcuts, and system usage.

macOS / Productivity / Other

McanKul/offpdf: A suite of 23 fully local, open-source PDF tools for editing, merging, converting, compressing, OCR, and more, with no uploads and complete offline support.

ddlmanus/MacOptimizer: A SwiftUI macOS optimization and application-management utility with resource monitoring, app removal, junk cleanup, and large-file discovery.

AeroSpace Guide: The complete guide to AeroSpace, an i3-like tiling window manager for macOS.

Hammerspoon: A macOS automation tool that bridges system APIs and Lua to orchestrate windows, applications, keyboard and mouse input, the clipboard, networks, and hardware events.

Cap: An open-source tool combining screen recording, screenshots, editing, and video hosting, with support for storing recordings in your own Google Drive or S3 bucket.

Inkdrop: A cross-platform Markdown note app for developers with end-to-end encrypted sync and note-context access for MCP clients such as Claude Code.

How things win: Steph Ango distills 80 strategies for building advantages, creating moats, and surviving over time from examples in biology and business.

Departure Mono: A retro pixel-style monospaced font by Helena Zhang, released for free under the SIL Open Font License.

Design bookmarks: Anthony Hobday’s topical collection of bookmarks for web and visual design references.

Arcotype: A curated directory for discovering less overused typefaces, all available under the SIL Open Font License.

Tuval: An open-source tool that turns an Android tablet into a graphics tablet, supporting pressure, tilt, hover, low-latency transport, and screen mirroring.

starcat-app/Starcat: A native macOS GitHub Stars manager that turns saved repositories into a personal knowledge base with tags, notes, full-text and semantic search, AI summaries, and more.

Metalterm: A native macOS terminal for Apple Silicon that renders through Metal 3 and provides command blocks, fast search, and a command palette.

missuo/herdrm: A native macOS console for herdr that brings together coding-agent status across local and remote devices and connects directly to each live terminal.

Writer: A free, open-source, lightweight, local-first Markdown editor for macOS with Mermaid, tables, HTML, YAML Frontmatter, and multi-window workspaces.

Pulse: A native macOS menu-bar stock ticker for quickly checking watchlist prices and market movements.

Normal Tools: A collection of 193 calculators, converters, checkers, generators, and hardware tests that run directly in the browser with no account or download.

Product Now: A daily curated discovery site for new digital products in AI, design, productivity, and related fields.

Look: A keyboard-first macOS launcher focused on quickly performing local actions.

@@ for Mac: A native tool for invoking an AI agent from any text field on macOS.

byrencheema/airstats: A low-overhead macOS menu-bar system monitor showing CPU, memory, GPU, network, disk, battery, and temperature data.

Learning more about Claude’s mathematical capabilities — Anthropic describes how an unreleased Claude research model raised the verified lower bound on the proportion of Riemann zeta zeros satisfying the hypothesis from 41.6% to 67.2%, with a paper and Lean formalization.

Dyna-2: A 1-Million-Hour Scaling Law for World-Action Models — Dyna Robotics presents a world-action model pretrained on more than one million hours of human video and studies its scaling behavior on human and robot tasks.

[2608.09867] Stealing Reasoning Traces from Proprietary LLM APIs — demonstrates how reusing encrypted reasoning blocks across sessions may expose proprietary reasoning traces, and discusses prompt-injection risks and mitigations.

Introducing Unsloth Desktop — an open-source desktop application for Windows, macOS, and Linux that runs, trains, exports, and deploys several kinds of AI models locally.

MiniMaxAI/MiniMax-Music3 · Hugging Face — MiniMax’s model and weights page for its music-generation model.

Training AI Scientists to Replicate Research · Inherent — introduces Faraday, a 27B AI research agent trained with long-horizon reinforcement learning to reproduce figures from research papers.

Measuring Autonomous AI Research — reports an evaluation of 18 frontier models across 153 autonomous runs on a nanoGPT optimizer speedrun task.

SlimeBoyOwO/LingChat — an AI-powered galgame chat application with expressive characters, a desktop pet, scheduling, and interactive story modules.

Mathematics / Research / Learning

Problems · TheoremDB — a database of mathematics and theoretical computer science problems filterable by field, solution status, and Lean verification status.

[2606.03953] Introduction to stochastic gradient methods — a systematic introduction to deterministic and stochastic first-order optimization, covering smoothness, convexity, the PL condition, convergence rates, and variance reduction.

ComfyResearch: See How Learning Happens — represents data, models, training, and observations as composable graph nodes so researchers can visualize and compare neural-network learning dynamics.

Hadamard Matrix of Order 668 | Epoch AI — a FrontierMath problem on constructing a Hadamard matrix of order 668; the page records a candidate solution developed collaboratively by three researchers and Claude.

Matrix Calculus — an online symbolic tool that differentiates vector, matrix, and tensor expressions and exports results to Python or LaTeX.

• [2010.03313] A Simple and Efficient Tensor Calculus for Machine Learning — proposes an Einstein-notation-based tensor differentiation algorithm designed for efficient higher-order derivatives and automatic-differentiation representations.
• Computing Higher Order Derivatives of Matrix and Tensor Expressions — presents a framework for directly computing higher-order matrix and tensor derivatives using symbolic, forward, and reverse automatic differentiation.

GENO — describes optimization problems in a modeling language and automatically generates Python solvers for CPUs or GPUs.

• [1905.13587] GENO — GENeric Optimization for Classical Machine Learning — introduces the GENO framework for generating specialized solvers from declarative problem descriptions and evaluates it on classical machine-learning problems.
• [2203.16340] Optimization for Classical Machine Learning Problems on the GPU — extends GENO to constrained GPU optimization and compares solver performance across classical machine-learning problems.

How to keep thinking — considers how frequently reviewing LLM output can affect habits of slow thought, and recommends writing and reading dense books to preserve deeper thinking.

Free eBooks | Project Gutenberg — a free ebook library focused primarily on older works whose U.S. copyrights have expired.

Sakai Association of Science Summer Training 2026 — the 2026 summer training course site for publishing preparatory material, lecture notes, and assignments.

ChapterPal — AI-assisted reading and note-taking — an AI-assisted reading and note-taking tool with contextual Q&A, active recall, and incremental reading.

Math Translations — a community for openly preserving, sharing, and collaboratively maintaining Chinese translations of mathematics texts, lecture notes, and terminology.

Algorithm Wiki — a knowledge base that organizes algorithm material by field, problem, algorithm family, and variant.

Programming / Systems / AI Tools

jcode — an open-source terminal coding agent written in Rust, emphasizing low resource usage, parallel sessions, and an auditable, customizable runtime.

QwenLM/Qwen-MM-Plugins — multimodal plugins for Qwen models and multiple agent frameworks, covering images, video, OCR, search, media editing, CAD, and more.

Optimising fp32 GEMM on A100 — progressively explains and benchmarks Tensor Core, PTX, shared-memory, asynchronous-loading, and pipeline optimizations for fp32 GEMM on NVIDIA A100 GPUs.

lidge-jun/opencodex — a local proxy that translates Codex and Claude Code client protocols to multiple LLM providers, with streaming, tool calls, and account routing.

duolahypercho/codex-router — a local router that connects external Anthropic, Kimi, DeepSeek, and other models to Codex’s native model-selection flow.

PI from Scratch — a Chinese tutorial for building a personal pi-agent from scratch; the page is currently unavailable, so this description relies only on its public title.

lassejlv/gpui-starter — a minimal Rust desktop application template built with Zed’s GPUI, including a basic theme and interaction example.

Beautiful UI — Crafted primitives for AI-native interfaces — copyable interface primitives for AI-native applications, covering streaming states, approvals, tool calls, tasks, diffs, tables, and chat interactions.

raullenchai/Rapid-MLX — a local model-serving engine for Apple Silicon with OpenAI- and Anthropic-compatible APIs for coding agents and other clients.

Waku — one native app for all your coding agents — a native Rust and GPUI application that manages sessions, tool activity, and Git checkpoints for multiple coding agents in one interface.

pytorch-training-recipe · ml-training · pytorch-lightning — three agent skills for training workflows, covering reproducible PyTorch training, general machine-learning checks, and PyTorch Lightning project organization.

DeepSeek Harness — DeepSeek’s open-source agent runtime with an everything-is-a-plugin architecture, currently in developer preview.

Package Catalog · Pi — a catalog of npm-distributed extensions, skills, prompt templates, and themes for the Pi coding agent.

zhu1090093659/dsh-web-ui — a community collection of plugins and skins for the DeepSeek Harness Web UI, including a task board, Git graph, and mobile remote interface.

kenn-io/agentsview — a local-first tool for searching and analyzing coding-agent sessions and token usage.

Zeron — monitors and controls local Claude Code, Codex, Grok, Hermes, and Pi agents from another device in real time.

DannyMac180/grok-plugin — uses MCP and xAI OAuth to let Claude Code or Codex delegate tasks and code reviews to Grok.

xiaobright/dsh-anchored-standard — an experimental community DeepSeek Harness preset that stages the tool catalog to shape initial session behavior before enabling the full toolset.

dshfind — a Chinese community that aggregates DeepSeek Harness plugins and offers plugin metrics, courses, and paper explainers.

blader/humanizer — an agent skill for reducing common signs of AI-generated prose.

Maximizing the value of your Claude Code sessions — Anthropic’s guidance on getting more value from Claude Code sessions, with an emphasis on using tokens effectively.

Home | mise-en-place — a polyglot CLI for managing project tool versions, environment variables, and tasks.

GPUI Component — a component library for building cross-platform desktop applications with Rust and GPUI.

Welcome to Omarchy! — The Omarchy Manual — the official manual for Omarchy, a Linux distribution built on Arch, Hyprland, and Quickshell.

ast-grep — a fast structural code search, linting, and batch-rewrite tool for multiple programming languages.

macOS / Productivity / Other

Tinycast — a lightweight native macOS launcher built with SwiftUI and AppKit, with app launching, hotkeys, clipboard history, and snippets.

History of Market — interactive charts covering the long-term performance, returns, valuations, and drawdowns of the S&P 500, Nasdaq, and other U.S. markets.

VibeCleaner — a native macOS utility that scans and classifies Xcode, package-manager, Docker, Ollama, and other development caches by cleanup safety.

Student Education Discount Directory — a directory of student discounts and education-email verification links for software, shopping, entertainment, and AI tools.

metedata/Remarc — captures feedback on a Mac through text, screenshots, web elements, or voice and passes it to an agent over MCP.

Savant — a note-taking application for Mac and iPhone, currently in early alpha.

Moeka — a Markdown editor focused on a fluid, high-performance editing experience.

Former OpenAI reasoning lead JT and Gemini pretraining lead Rohan on the future of continual learning — a conversation about continual learning, post-pretraining knowledge updates, and long-term model adaptability.

(Some of) The Models, They Just Don’t Want to Learn | Tilde — examines potential recursive depth and training difficulties in reasoning models, with the One Layer Deeper competition exploring the joint design of architectures, optimizers, and objectives.

Maple-Preview Model Card — DeepGrove — DeepGrove’s open-source 20B-A1B reasoning model with ternary weights, designed for fast on-device inference and adaptation.

Xiaomi-Robotics-1 — Xiaomi’s robot foundation model, pretrained on more than 100,000 hours of embodiment-agnostic manipulation trajectories, then aligned with target embodiments and instructions using real-robot data.

Lattice: an 8 MB static retriever that embeds Wikipedia in 7 minutes — an approximately 8 MB static retrieval model and pure-Rust runtime that can embed 6.4 million English Wikipedia articles in about seven minutes on an 8-core M2 MacBook Air.

[2606.12683] From AGI to ASI — outlines four possible paths from AGI to ASI: scaling, paradigm shifts, recursive improvement, and large-scale multi-agent collectives.

What are AI robots? A complete introductory guide to 2026 — an introductory guide to AI robots, their core components, common types, and applications.

Evolution Strategies at the Hyperscale — introduces EGGROLL, which uses structured low-rank perturbations to increase the GPU arithmetic intensity of evolution strategies; the authors report roughly 100× higher throughput than naive ES at billion-parameter scale and large population sizes, with experiments on pure-integer RNN pretraining and RWKV reasoning post-training.

Mathematics / Research / Learning

The Daily Diff — a daily collection of worthwhile papers, technical articles, and discussions with recommendation signals and concise summaries.

TensorTonic — an interactive platform for learning machine learning by implementing algorithms, CUDA kernels, and landmark papers from scratch.

Interactive Machine Learning — Prof Richard Xu — an online course that explains machine-learning concepts and algorithms through interactive material.

Inference in Statistical Modelling and Machine Learning — the companion site for a textbook on inference in statistical modeling and machine learning, with a complete PDF, solutions, slides, and Python / R notebooks.

Inference Engineering — an interactive guide to AI inference spanning GPUs, CUDA kernels, quantization, KV caches, parallelism, and production deployment.

Kurate.org — AI Paper Rankings — a paper-ranking site where three LLMs jointly assess the potential impact of preprints in robotics, economics, physics, and other fields.

AI Settles a 25 Year-old Problem We Left Behind — discusses the binary integer least-squares problem underlying maximum-likelihood MIMO detection and an example of AI-assisted progress on related theory. Exact solution in general remains NP-hard; polynomial-time results usually refer to relaxations or approximate detectors. The target problem is:

JoseDeFreitas/awesome-youtubers — a curated list of technical YouTube channels covering web development, computer science, machine learning, game development, cybersecurity, and more.

Programming / Systems / AI Tools

jcode — an open-source terminal coding agent written in Rust, emphasizing low resource usage, parallel sessions, persistent long-running work, and auditability.

VisualTeX — an online tool for visually editing and previewing LaTeX formulas in real time.

Peezy — a terminal coding agent for real projects with a unified interface for switching among multiple models and OpenAI-compatible providers.

Diffs, from Pierre — an open-source code and diff rendering library built on Shiki, with split and unified views, annotations, conflict handling, and interactive accept-or-reject controls.

yetidevworks/drydock — a Rust TUI for checking uncommitted changes, unpushed commits, repositories behind their remotes, and unreleased versions across multiple repositories.

Canario — a macOS terminal built on the Rio engine, with spatial grouping, splits, a command palette, tab previews, and a global quick terminal.

Local Studio — a local AI workspace that brings models, inference runtimes, hardware monitoring, files, terminals, and agents into one interface, with unified access to vLLM, SGLang, MLX, and llama.cpp.

CNB — Tencent’s cloud-native development platform, managing build environments, caches, plugins, and elastic compute through an Everything as Code model.

SteveSuv/remix-words-funny — a full-stack English vocabulary-learning project built with React Router v7, tRPC, Drizzle, PostgreSQL, HeroUI, and Docker.

codecast-sh/codecast — records, searches, and remotely steers Claude Code, Codex, Cursor, and Gemini sessions, with team memory and line-level agent attribution.

macOS / Productivity / Other

Tahul/space-rabbit — removes macOS Spaces switching animations to reduce desktop-transition delays.

Vorssaint · vorssaint/vorssaint-utils — a free, open-source modular macOS menu-bar toolkit for sleep prevention, hardware monitoring, volume, networking, disks, and common system controls.

Side Screen — turns an Android tablet into an extended display for a Mac over USB-C or Wi-Fi, with hardware acceleration and touch input.

migsilva89/imark — a native macOS Markdown reader that refreshes as files are saved; annotations are stored directly as HTML comments in the Markdown file without uploading it to the cloud.

moonshotai/Kimi-K3 · Hugging Face — Moonshot’s Kimi K3 model weights and model card.

ali on X: “22580: From GPT2 to Kimi3, Explained” — a walkthrough of architecture evolution from GPT-2 to Kimi K3.

cneuralnetwork/smol-kimi-k3 — a ~49M-parameter Kimi-K3-style small model trainable on a single 8GB GPU.

cneuralnetwork/smol-glm-5.2 — a ~50M-parameter GLM-5.2 architecture study with MLA, DSA, IndexShare, sparse MoE, shared MTP, and Muon Split.

Exclusive | Nvidia Invests in Sutskever’s Safe Superintelligence AI Lab - WSJ — Nvidia invests in Ilya Sutskever’s SSI to expand compute reach and collaboration.

Compression and Intelligence — Ryan Greene — on the relationship among compression, intelligence, and LLMs.

Physics of Language Models — Allen Zhu et al.’s “physics of language models” series, explaining LLM behavior via architecture and training mechanisms.

Deconstructing Scaling Laws: Optimization, Architecture, and Data - Scientific Spaces — unpacks scaling laws through optimization, architecture, and data.

MinghuiChen43/awesome-deep-phenomena — a curated, continually updated list of interesting empirical deep-learning phenomena and papers.

AI Explorables | PAIR — interactive AI concept explorables from Google PAIR that visualize model mechanisms.

Looped Transformer | alphaXiv — a paper collection on looped Transformers.

Towards Looped Models Done Right — on designing and training looped / recurrent-style models properly.

Explorative Modeling — Unlocking a Third Pretraining Axis and End-to-End Generation | Alexi Gladstone — proposes explorative modeling as a third pretraining axis. Core idea sketch:

losses = []
for i in range(K):
    generation = model.generate()  # e.g., from a different random noise
    losses.append(loss_fn(generation, data))
min(losses).backward()  # only the best generation gets gradients

What optimization dynamics appear if $\min_{i} \mathcal{L}_{i}$ is generalized to an arbitrary $f(\mathcal{L}_{1},\dots,\mathcal{L}_{K})$?

[2604.22951] The Power of Power Law: Asymmetry Enables Compositional Reasoning — argues that the power-law asymmetry of natural language helps compositional reasoning and should not simply be flattened.

Enigma: Robots that feel right — Enigma on robots that understand their surroundings and act in ways that “feel right.”

maka-agent/maka-agent — a local-first AI desktop assistant.

ArXivIQ | Grigory Sapunov | Substack — hand-picked ML papers with AI-generated deep dives; daily TLDR and architecture notes.

Blog - Ulam · Research - Ulam — AI research notes and projects such as DeepAlgebra.

Home | CSE 234 — course page for CSE 234 Machine Learning Systems.

Mathematics / Research / Learning

Mathematics in the age of AI — Terence Tao’s slides on mathematics in the age of AI (ICM 2026 context).

What Is the Inverse Covariance Matrix? — explains the precision (inverse covariance) matrix from statistical and geometric viewpoints.

[2607.21716] A Defense of the Quadratic Model — defends the quadratic model in optimization theory as a tradeoff between tractability and real loss landscapes.

AI Agents for the Working Mathematician — practical AI-agent workflows for working mathematicians.

Prompting Toward a Conjecture - Guanyang Wang — how one extra sentence about searching the literature changed GPT-5.6 Sol Ultra’s success rate on Feige’s $1/e$ conjecture.

Generating Proofs with LLMs and Completing Formal Verification - Guanyang Wang — LLM-generated proofs plus Lean formal verification, illustrated on the binary-matrix case of the 1997 KTV conjecture.

Ten advances in mathematics and theoretical computer science | OpenAI — OpenAI’s summary of ten advances in mathematics and theoretical computer science.

UnsolvedMath — a repository of open mathematics problems.

Home | Open Problem Garden — a directory and community garden of open problems in mathematics.

Optimization Online — an eprint repository for optimization and related topics.

Trouble in High-Dimensional Land - Dibya’s Blog — counterintuitive phenomena in high-dimensional geometry and probability.

faridrashidi/cnsplots — a toolkit for Cell / Nature / Science-style publication-quality plots.

Programming / Systems / AI Tools

Herdr: one terminal for the whole herd — an agent multiplexer: one terminal for a whole herd of agents.

Shuffle - A fast, native file manager for macOS — a free open-source native macOS file manager (Rust + GPU rendering) with dual panes, tabs, command palette, and terminal mode.

Tinycast — a tiny, native macOS launcher — a lightweight native macOS launcher with fuzzy search, calculator, clipboard history, and global hotkeys.

Rust Vs C++ - LLM coded model conversion from PyTorch — LLM-written PyTorch model-conversion logic compared side by side in Rust and C++.

icebird1998/drawio-scientific-illustrator — step-by-step scientific illustration via MCP control of a live draw.io canvas.

zeronsh/chat — a unified AI chat app spanning Claude, OpenAI, Gemini, and more.

terminal-browser — a real browser that runs inside the terminal.

Codex Reset — Last Reset Tracker & Live Forecast — tracks Codex usage-reset times with 24h / 48h forecasts.

GPU Glossary — Modal’s glossary of GPU terms.

ascend-ai-coding/awesome-ascend-skills — a Huawei Ascend NPU knowledge base organized as distributed Agent Skills.

Doubao Search Custom — Volcengine Doubao Search docs; 500 free calls per month.

iv-org/invidious — an alternative YouTube front end.

Design / Writing / Other

://argon.red — a designer’s blog; related discussion on X from Alpha Argon.

Soofi S — Sovereign German-English Foundation Model — Soofi Project’s pretraining report for a German–English foundation model.

Learning from experience instead of curated datasets · Oak Lab — a discussion of how agents can learn from experience rather than curated datasets.

Inkling: Our Open-Weights Model — an open-weights model released by Thinking Machines Lab.

Kimi K3: A New Frontier in Intelligence — an introduction and reading of Kimi K3.

A four-hour interview with Physical Intelligence’s Ke Liyiming — a conversation about PI’s open-model research, the evolution of robotics, and its major participants.

Have Chinese AI Models Caught Up to the US Frontier? — uses benchmarks and capability-gap estimates to ask whether Chinese models have caught the US frontier.

Agent Swarms and the New Economics of Models — Cursor’s discussion of agent swarms and their model economics.

From Speaking to Creating: Seed Audio 1.0 — ByteDance Seed Audio 1.0, an audio creation model release note.

[2607.10252] One Token Is Enough: Fingerprinting and Verifying Large Language Models from Single-Token Output Distributions — fingerprints and verifies LLMs from single-token output distributions, useful when the serving chain is opaque.

Introducing Laguna S 2.1 — Poolside’s Laguna S 2.1 release, emphasizing longer-horizon work and more effective use of reasoning.

Macaron-V1-Preview: 749B MoL Agent Model post-trained from GLM5.1 — a 749B MoL agent model preview post-trained from GLM5.1.

[2507.09369] A Taxonomy of Omnicidal Futures Involving Artificial Intelligence — a taxonomy and examples of AI-related futures in which nearly all humans are killed; framed as possibilities, not inevitabilities.

Scaling Video Pretraining with Imagination Models — Induction Labs pretrained Photon-1 from scratch on 575 million frames of computer-use data, using autoregressive training in latent space.

[2607.20548] SOAP, Muon, and Beyond: Pushing LLM Pretraining Scales · NVIDIA-NeMo/Emerging-Optimizers — adapts and strengthens higher-order optimizers such as SOAP and Muon for large-scale pretraining, with related NVIDIA NeMo implementations.

How to train a frontier-level world model · next-state/open-dreamer — an open-source Dreamer implementation in JAX.

Causal Video Models Are Data-Efficient Robot Policy Learners — Rhoda AI’s DVA reformulates robot policies as video generation for better data efficiency and long-context behavior.

What is the purpose of interpretability? — asks what interpretability research is for and what problems it should serve.

Mathematics / Research / Learning

DSPIllustrations.com — interactive illustrations for understanding digital signal processing.

Wireless Pi — learning material on wireless communications and signal processing.

intro-to-dl — an introductory deep-learning course.

[1708.09093] Memories of a Theoretical Physicist — Joe Polchinski’s memoir-like notes as a theoretical physicist.

[2607.11664] Lecture Notes: Convex Optimization — lecture notes from a Graz University of Technology convex-optimization course, covering preliminaries, existence, and algorithmic basics.

Discussion of a three-dimensional map related to the Jacobian Conjecture — the original post gives a map from $\mathbb{C}^3$ to $\mathbb{C}^3$ and claims it is a counterexample; the conclusion should still be judged against a formal paper and peer review.

PyTorch internals — a guide to PyTorch’s internal mechanisms.

OpenResearch — an entry point for research resources and tools.

Byrne’s Euclid — an online edition of Oliver Byrne’s colorful rendering of Euclid’s Elements.

An introduction to the 2026 Fields Medals — a discussion of laureates Yu Deng and Hong Wang and their work.

Python Tutorials – Real Python — high-quality Python tutorials and practical articles.

[2607.20732] librla: Randomized Linear Algebra Library — a randomized linear-algebra library aimed at intermediate-sized matrices (up to roughly $10^4$ dimensions).

Programming / Systems / AI Tools

Synara — a development tool integrating Claude Code, Codex, Gemini, and Cursor.

ChatCut ChatGPT/Codex Plugin — a ChatGPT / Codex plugin for trimming and organizing conversation context.

xai-org/grok-build — an open-source coding-agent harness and full-screen TUI.

ZacharyZhang-NY/Kigi-CLI — an unofficial Rust community build of Kimi Code CLI based on Grok Build.

tldraw offline · nn-digits.tldraw example — an AI-operable desktop whiteboard.

petersolopov/yace — a browser code editor with plugin support in under 2 KB.

ayghri/i-have-adhd — an Agent Skill that constrains Claude Code’s output and brings the answer forward.

mesamirh/MovieBox-Tui — a MovieBox terminal interface written in Rust.

ChatGPT Table of Contents — a browser extension that generates a table of contents for ChatGPT conversations.

marcelroed/gigatoken — a language-model tokenizer designed for GB/s throughput.

Tangent — a browser centered on workflows and spatial organization.

Why I chose Tauri, and what it cost me — Tabularis on the experience and cost of choosing Tauri.

CaliCastle/skills — its Signal Geometry Skill turns one concept into a sparse abstract illustration or poster through precise geometry, low-contrast matte fields, and restrained spatial relationships.

nexu-io/codex-slides — an open-source AI slide studio inside Codex with support for research, outlining, rendering, editing, presenting, and export.

Minibase — converts web pages to Markdown for Claude, ChatGPT, and Obsidian.

l0ng-ai/tty7 — a terminal workbench built in Rust with Zed’s GPUI, integrating shells, persistent sessions, SSH, and coding agents; the project reports roughly twice the throughput of Alacritty, Ghostty, or Kitty.

Design / Writing / Creative Work

NameThat — a visual dictionary of UI elements, useful for identifying unfamiliar patterns and describing design needs to AI more precisely.

Arjun Kocher — a portfolio website.

Making Software: How to make a font — a tutorial on making a font.

Vercel Design System · Dark — a complete set of guidelines for color tokens, typography, spacing, radii, components, motion, and writing. Its central principle is that color should communicate, not merely decorate.

Feldar · Marker · Ornata · Ishmael — apps focused on concentration, thinking, or long-form writing.

Attune — a personal computing / writing app oriented toward focus and rhythm.

pixgel — a liquid pixel-art editor with smooth SVG export.

blakeshao/a-psychos-gd-tool — a node-based graphic-design tool.

You’re relatively right! — a heatmap of writing-style similarity across LLMs, built from wording alone.

Sticker Forge — an interactive sticker maker.

Canvas UI — a creative-canvas and WebGL component library.

Design inspiration and reference indexes:

• posts.design — social-post design
• noiced.com — daily web-design inspiration
• mnmm.xyz — a directory of minimalist websites
• ogpedia.xyz — OG image design
• deck.gallery — presentation design
• logosystem.co — logo-design examples
• visualjournal.it — branding and editorial projects
• brandguidelines.net — a library of brand guidelines
• recent.design — recent design work

News / Life / Other

赛博鸡蛋 — digital products, deals, and offers.

Electroagenda — an electronic-music event calendar.

jobleap — an online job-application information tool.

Eli Bendersky’s website — writing on programming, computer science, and technology.

Scenic Relaxation · Jacob Adler · Relaxation Film — channels featuring landscapes and relaxation films.

There Is No Avenue of Communication Between You and God — an essay about the absence of a direct channel of communication between people and God.

GPT-5.6: Frontier intelligence that scales with your ambition | OpenAI — OpenAI’s announcement of its next generation of frontier models.

Ethan Knight on X: GPT-5.6 Sol Ultra and the Cycle Double Cover Conjecture — reports that 64 subagents produced a proof of the Cycle Double Cover conjecture in about an hour; proof · prompt.

A practical workflow for proof search with multiple agents:

1. Define and formalize the problem precisely: input objects, permitted structures, output format, invariants, boundary cases, and acceptance criteria.
2. Specify both success and failure: which results count as correct, and which do not?
3. For an open problem, assume that a complete affirmative proof exists. This can bias the search toward accepting existing material as correct, so pair it with adversarial review.
4. Have agents explore independent directions. Avoid suggesting a preferred approach unless there is a definite reason to trust it.
5. Track each agent’s progress separately and standardize the output format.
6. Require verifiable deliverables.
7. Have a lead agent synthesize the results, maintain the overall state, and keep iterating through failed attempts.
8. Restrict searches to foundational material, excluding work directly related to the target problem.

The corresponding principles:

1. Specification before reasoning: define the objective, objects, and failure conditions precisely.
2. Diversity before consensus: preserve independent approaches before sharing and combining results.
3. Mechanisms, not narratives: accept only verifiable lemmas, constructions, equations, and counterexamples.
4. Optimistic exploration, adversarial verification: explore boldly and scrutinize the results rigorously.
5. Manage the search, not just the answers: record approaches, identify duplication, close dead ends, and allocate computing resources dynamically.

Prompting | ChatGPT Learn — official ChatGPT documentation on prompting.

30 papers · The reading list Ilya Sutskever gave John Carmack — a website collecting the 30 papers Ilya Sutskever recommended to John Carmack.

AI 2040: Plan A — material outlining a roadmap for AI through 2040.

[2607.01233] Measuring the Gap Between Human and LLM Research Ideas — a paper quantifying the gap between research ideas proposed by humans and large language models.

Optimization / Machine Learning / Data

[2607.04033] OmniOpt: Taxonomy, Geometry, and Benchmarking of Modern Optimizers — a taxonomy, geometric perspective, and benchmarking framework for modern optimizers.

J-lens CKA explorer — an interactive visualization of centered kernel alignment and representation similarity.

LAION Dataset Explorer — a browser for exploring LAION data in embedding space.

PolymathicAI/the_well / The Well — a collection of physics simulation datasets totaling about 15 TB.

Mathematics / Theory / Open Problems

AI Math Open Problems — open problems at the intersection of AI and mathematics, covering matrix integral inequalities, compressed convex functional calculus, lower bounds for tensor methods in monotone variational inequalities, information geometry, graph theory, and other topics.

Frank Nielsen | Information Geometry, divergences, and diversities — a resource map for information geometry, divergences, and diversity measures.

Algorithms by Jeff Erickson — Jeff Erickson’s algorithms textbook, freely available online.

thatmagicalcat/txm — a mathematics rendering engine for the terminal.

Programming / Systems / Tools

The CUDA Handbook — Chapters — chapters from The CUDA Handbook.

Cloudflare Drop — upload files to Cloudflare without signing in to deploy a temporary website that lasts about an hour.

Mounting Cloudflare R2 into macOS Finder — practical notes on mounting Cloudflare R2 in macOS Finder.

Harbofly · understand your storage, recover space safely — a tool for analyzing disk usage and safely reclaiming storage.

johnfactotum/foliate — an ebook reader for Linux with a clean interface.

Philosophy / Publishing / Other

History of Philosophy - Summarized & Visualized — an interactive visualization and summary of the history of philosophy.

Explore · PubPub — a directory of open publishing projects and knowledge communities.

Manggo — a new translation tool from the author of Pot after Pot entered maintenance mode. The original Pot is still useful migration context.

opendatalab/Sciverse-Agent-Tools — agent tool schemas and SDKs for the Sciverse Open Platform, organizing retrieval capabilities into standard tools that LLM agents can call.

Terax — a terminal-first AI-native development workspace that brings AI development workflows back into the terminal.

Gitlawb/zero — a local, autonomous, controllable coding agent built around the idea that your model, machine, and rules are yours.

Tiktokenizer — an online tokenizer inspection tool for quickly seeing how text is tokenized.

偷懒工具 — a collection of small online tools for everyday productivity.

UniClipboard — a local-first, P2P, end-to-end encrypted cross-device clipboard sync tool with no account or cloud relay required.

Machine Learning / Optimization / Courses

Random Reshuffling Dominates Stochastic Gradient Descent — proves that random reshuffling with suitable step sizes outperforms SGD after any finite number of epochs in smooth convex optimization, resolving a long-standing open question.

Orca: The World is in Your Mind — a world model paper page focused on internal world representations and predictive ability.

Reinforcement Learning of Large Language Models — course material on reinforcement learning for large language models.

DSC 243: Convex Quadratics — lecture notes on convex quadratic functions from an advanced optimization course.

Pen and Paper Exercises in Machine Learning / GitHub — machine-learning exercises suited for conceptual derivation and exam-style practice.

Presentations — Michael Gutmann’s presentations on statistical machine learning.

Machine Learning Visualized — visual material for machine-learning concepts.

IClimbTrees — a collection of AI and machine-learning courses.

Signal / Radar / Audio

How I built a mmWave material classification radar — a personal project write-up on mmWave material classification radar, including hardware, signals, and experiments.

AudioMass — an online audio editor for lightweight editing and waveform processing.

Computer Fundamentals / Theory / Engineering Articles

CS-Fundamentals — interview-prep material covering DSA, computer networks, DBMS / SQL, OOP, operating systems, system design, and software engineering.

Respect the P v NP Problem — a Computational Complexity blog post on the P vs NP problem.

(Quite) A Few Words About Async — a long article on async programming, useful for revisiting the design motivations and sources of complexity behind async.

Design / Illustration / Image Resources

Open Peeps — a hand-drawn people illustration library.

Old Book Illustrations — a public-domain old-book illustration archive.

Online image vectorizer — an online bitmap-to-vector tool.

Wenyang — a reference site for traditional Chinese patterns.

Sources / News / Life Tools

Hatnote Listen to Wikipedia — turns Wikipedia editing activity into real-time sound and visual feedback.

Data Science Weekly Newsletter — a data science weekly newsletter.

FiNews — a daily US stock-market news digest.

Showpaw / 妙搜云 / 小云搜索 / 小狐搜搜 — search tools for media, study materials, software, and public cloud-drive resources.

Lie-flat city guide — a city-selection reference for freelancers and remote workers.

folia-major — a local music / Navidrome / third-party NetEase Cloud Music player project focused on vivid lyric animations.

Depo | AI Harness tools / Depo - Curated Tools, Skills & Resources — a tool directory for AI agents and developer workflows.

rebel0789/codexpro — connects ChatGPT Developer Mode to a local coding agent through MCP.

Waishnav/devspace — a project for bringing ChatGPT into a Codex-like development space.

caezium/Burrow — a native macOS GUI for Mole CLI, covering cleanup, uninstalling, optimization, disk analysis, and MCP.

loops! | Pre-built agent loops — a collection of prebuilt agent loops.

Oak — an AI knowledge base for collaboration between people and agents.

JerryZLiu/Dayflow — a local-first automatic work log that organizes screen activity into a timeline.

Aside | The browser built to do real work for you — a browser focused on agentic work.

HuntScreens — a discovery site for new AI tools.

Codex Radar — an aggregator for Codex-related updates.

Previewing GPT-5.6 Sol: a next-generation model | OpenAI — OpenAI’s model preview article.

Agent Mail — an email product built for agents by QQ Mail.

Programming / Open Source Tools / Systems

Lore — a next-generation open-source version control system.

Otty — a native terminal application.

renatoworks/oh-my-reddit — a terminal tool for reading Reddit threads.

l0ng-ai/papr — a fast, native desktop RSS reader.

ReLaTeX - Copy & Render AI Equations — a browser extension for copying and rendering AI-generated equations.

cool-japan/trustformers — Rust-based Hugging Face Transformers inference and deployment work.

baidu/Unlimited-OCR — a long-document OCR / parsing project.

Blind 75 - LeetCode Problems with Video Solutions — the Blind 75 problem list with video explanations.

Mathematics / Machine Learning / Vision

Michael-Jetson/Robotics_Tutorial — systematic robotics teaching material covering math foundations, C++, planning and control, and SLAM.

Algebrica | A Mathematical Knowledge Base / antoniolupetti/algebrica — a mathematical knowledge-base project focused on clear structure.

Taxonomy of Principal Distances & Divergences — a taxonomy of common distances and divergences.

Natural Number Game — a Lean-style proof game for natural numbers.

A Self-Learner’s Roadmap Through 30+ Great Math Books — a self-study roadmap through mathematics books.

How To Scale Your Model — a JAX / ML scaling textbook.

Alisa’s book of LLMs / math notes — collections of LLM and math notes.

Deep-ML | Practice Machine Learning — a machine-learning practice platform.

Foundations of Computer Vision — MIT’s computer vision textbook.

deep Manifold on X: “Neural Network Learnability and Discoverability” — a thread on neural-network learnability and discoverability.

Among all branches of mathematics, complex analysis is closest to art — an essay on learning and appreciating complex analysis.

LLM Visualization — an interactive visualization of LLM architecture.

MLE is not intuitive — an article on the intuition behind maximum likelihood estimation.

plugyawn/Megaprop — an implementation of a preconditioned optimizer for MoE training, supporting MuP, FSDP Muon, and Megatron-LM.

Product / Design / Reading

Product Design Psychology by Wouter de Bres — resources on product design psychology.

Notes on the Industry Job Search — notes on industry job-search experience.

不想看片 / kanpianer/bxkp.org — a media-resource navigation site and its open-source repository.

Orthogonal — Greg Egan — the entry point for Greg Egan’s Orthogonal trilogy.

4K original disc downloads — a 4K original-disc resource site.

Codex Reset Radar — a small tool for tracking Codex reset rhythms.

Making Software — essays on software construction, engineering judgment, and product practice.

Alignment Science Blog — Anthropic’s alignment science blog.

galilai-group/stable-worldmodel — a reproducible experimental platform for world model research and evaluation.

WeightWatcher — a deep-learning model diagnostic tool that does not require training data.

NEKOparapa/AiNiee — a one-click AI translation tool for complex long-form content such as games, novels, subtitles, PDFs, and Word documents.

Building a hill-climbing machine — Microsoft AI’s launch article for the MAI model series.

arXiv 2605.23901: LLMs as Noisy Channels — a Seed team paper that views model capacity and scaling laws through Shannon’s noisy-channel lens.

Is Frontier Asynchronous RL Solved? — Luke Huang’s overview of frontier asynchronous reinforcement learning.

On the Legacy of Linear Transformers in Positional Embeddings — a look back at positional embeddings from the perspective of linear transformers.

arXiv 2605.27365: LocateAnything — a vision-language grounding method with parallel box decoding.

asgeirtj/system_prompts_leaks — a collection of system prompt snapshots from various AI tools.

Phistory / WEIFENG2333/phistory — an archive of version diffs for Agent CLI system prompts.

iAmCorey/kooky — a local terminal for AI coding, with workspace sidebars, split panes, and agent launching.

CarSmallGuo/deveco-code — a DevEco Code related project.

Programming / Systems / Developer Tools

microsoft/coreutils — a Coreutils installation and packaging project for Windows.

Introduction - Rust for Python Programmers — an introduction to Rust for Python programmers.

Grit: rewriting Git in Rust with agents — a GitButler article on rewriting Git with Rust and agents.

nkzw-tech/codiff — a fast local diff viewer.

lbq110/weread-exporter — a WeRead full-book exporter based on Playwright and Canvas hooks.

qianchd/SnapTeX — a LaTeX / equation screenshot tool.

philexohf/transformer-zh-en — a native PyTorch Transformer implementation for Chinese-English translation.

Mathematics / Statistics / Signal Processing

Statistics self-study guide — a self-study roadmap and resource list for statistics.

Home · Dandelion Book Series — the entry point for the NNDL / Dandelion book series.

Math Curve Loaders — a collection of mathematical curve loading animations.

Keith Conrad’s Home Page / Bob Gardner’s Courses — math course and lecture-note resources.

Towards Compositional Steepest Descent — Tilde’s article on Compositional Muon and steepest descent.

How to understand MCMC intuitively? — an intuitive explanation of MCMC.

Deep Learning Notes — Chinese deep-learning notes.

xiyin137/QFT — a repository related to quantum field theory.

GNU Radio tutorial / btfz-sdr/gnuradio_demo — GNU Radio tutorials and demo projects.

Is it reasonable to love algebra and dislike probability, especially mathematical statistics? — a discussion about learning preferences across algebra, probability, and statistics.

Fermat’s Library — annotated classic papers and mathematical texts.

Academic Writing / Papers / Knowledge Bases

Vide — the VIDE project from PASCAL Lab.

LocoWiki — an online wiki / knowledge-base project.

Wikigraph — a tool for browsing Wiki content as a graph.

Academic paper writing methods and techniques for machine translation — lecture slides on writing machine translation papers.

How the Open Knowledge Format can improve data sharing — a Google Cloud article on Open Knowledge Format and data sharing.

PP-OCRv6 — PaddleOCR v6 model collection.

Design / Visuals / Reading

Civilization VI Encyclopedia — a Civilization VI encyclopedia.

OPENAI(ChatGPT) SMS price comparison — price comparison for ChatGPT / OpenAI SMS receiving services.

Browse the Collection — Storied Colors — a color reference site organized around historical stories.

Ogden’s Basic English · 850-word handbook — learning resources for Basic English’s 850-word vocabulary.

CS294/194-280 Advanced Large Language Model Agents — UC Berkeley’s advanced LLM Agent course, useful for systematically learning agent architectures, evaluation, and practice.

Large Language Model Agents — an LLM Agent course homepage covering tool use, planning, memory, multi-agent systems, and other foundations.

9Router - Free AI Router | Smart Fallback for Claude, Codex & More — a large-model gateway integrating RTK and Caveman Mode to save tokens.

AI 要跨越盧比孔河了嗎？自我成長的 AI 離我們多遠？ - YouTube — Professor Hung-yi Lee’s introduction to AI self-improvement.

WebBridge - Let Kimi Agent Drive Your Browser | Kimi — Kimi’s browser-control plugin.

memohai/Memoh: ✨ Self hosted, always-on AI agent platform run in containers. Create multiple bots with long memory, and connect them to Telegram, Discord, Feishu(Lark), Matrix, etc. — a self-hosted, always-on AI agent platform with long memory and IM integrations.

yetone/native-feel-skill: An Agent Skill for designing cross-platform desktop apps that feel native — distilled from Raycast’s 2.0 deep-dive and reverse engineering of Raycast Beta.app. Eight architectural tenets, four-layer architecture, WebKit/WebView2 survival guide, 75-item ship audit. — an Agent Skill for cross-platform desktop apps with a native feel, useful as a design and release checklist.

lishix520/academic-paper-skills: Systematic framework for planning and writing academic papers using Claude Code. Includes strategist (planning) and composer (writing) skills with quality checkpoints. — a set of paper planning and writing skills that breaks topic selection, structure, and drafting into checkable workflows.

a9gent/mindfs: Access your personal AI agents and workstation data anywhere, anytime through MindFS. — exposes personal agents and workstation data through a remotely accessible filesystem interface.

Mousedo — The most powerful computer-use agent for Mac — a Mac computer-use agent with cursor-based clicking, typing, app switching, and related actions.

Helio — the AI-native team workspace — an AI-native workspace for team collaboration.

Hive — 在浏览器里造一座 Agent 蜂巢 — a browser-based multi-agent workspace for coordinating multiple agents.

Cumora — Where agent teams gather — a product homepage for agent-team collaboration and orchestration.

AIHOT — an AI news aggregator that curates models, products, papers, tips, and other high-value content.

nexu-io/html-anything: ✨ The agentic HTML editor — your local AI agent writes the HTML, you ship it. 🚀 75 Skills × 9 Surfaces (magazine · deck · poster · XHS / tweet · prototype · data report · Hyperframes) 🛡️ Sandboxed preview · 📤 1-click to WeChat / X / Zhihu / HTML / PNG 🔑 Zero API key — Claude Code / Cursor / Codex / Gemini / Copilot / OpenCode / Qwen / Aider. — a local agent-driven HTML creation tool covering posters, cards, decks, prototypes, and data reports.

google/agents-cli: The CLI and skills that turn any coding assistant into an expert at creating, evaluating, and deploying AI agents on Google Cloud. — Google Cloud CLI and skills for creating, evaluating, and deploying agents.

An OpenAI model has disproved a central conjecture in discrete geometry | OpenAI — OpenAI’s case study on a model disproving a central conjecture in discrete geometry.

Open-Less/openless: Hold a key, speak, release — AI-polished text appears at your cursor in any app. Open-source voice input for macOS & Windows. (按住快捷键说话，松开即得润色后的文字) — an open-source voice input tool: hold a hotkey, speak, release, and polished text is inserted at the cursor.

FuJacob/tabby: tabby is local AI autocomplete for your entire Mac. Open source. On device. Everywhere you type. — local AI autocomplete across all typing contexts on macOS.

Inference Infrastructure — a long article on LLM inference infrastructure, useful for sorting out server-side inference, scheduling, and performance optimization.

Dictation354/paper-fetch-skill: Fetch papers as agent-ready markdown — DOI/URL/title in, structured full text out. CLI · MCP · Skill. — fetches papers into agent-ready structured Markdown from DOI, URL, or title.

Reading / Ebooks / Writing

双语电子书翻译器 — Philo Li — an ebook translation tool supporting EPUB and PDF.

yihong0618/bilingual_book_maker: Make bilingual epub books Using AI translate — another bilingual ebook maker, focused on EPUB.

Standard Ebooks: Free and liberated ebooks, carefully produced for the true book lover — a free ebook site with carefully produced editions of classics in multiple formats.

Welcome to the Purdue OWL - Purdue OWL® - Purdue University — a writing-resource site with guides and formatting rules, especially useful for academic writing.

📚 电子书下载宝库 - Ebook Treasure Chest — an ebook download index collecting many Chinese ebook resources.

GitHub: jbiaojerry/ebook-treasure-chest: 欢迎来到电子书下载宝库，一个汇聚了各类电子书下载链接的地方。无论你是喜欢阅读经典文学、经管励志、终身学习、职场创业、技术手册还是其他类型的书籍，这里都能满足你的需求。 该库涵盖了帆书app(原樊登读书)、微信读书、京东读书、喜马拉雅等读书app的大部分电子书。 — companion GitHub repository.

SaberOnGo/public-domain-books-translation: Public domain books translation project — a public-domain book translation project, useful if you care about bilingual classics and republication workflows.

macOS / Productivity / Terminal

Espanso - A Privacy-first, Cross-platform Text Expander — a very useful text expander that replaces keywords automatically, supports shell commands, and works cross-platform.

Obsidian Community — Obsidian’s new plugin community.

Genymobile/scrcpy: Display and control your Android device — control Android devices from a computer.

pranshuparmar/witr: Why is this running? — a TUI for seeing why a process is running.

Lakr233/SafariYYDS — Safari enhancement project, useful for collecting Safari ecosystem tools.

ThaddeusJiang/SafariYYDS-js — JavaScript version or related implementation of SafariYYDS.

rockorager/comview: the best diff viewer ever made — a command-line diff viewer for more comfortable code review.

Supermac - Superpowers for your Mac — a native Mac launcher and productivity tool focused on local-first behavior, privacy, and fast search.

Shaihai Studio — ∫aihai Studio’s site, with science-popularization and experimental video projects.

Academic / Math / Engineering Resources

zhuyidao.com — an “attention calculator” that can automatically notice integral constructions such as $\pi > 3.14$.

SciPy FFT vs NumPy FFT: 2.3x Speed Gap at 50kHz – TildAlice — SciPy’s FFT module can be much faster than NumPy’s, especially at larger scales.

liuskywalkerjskd/Control-Note: A concise and intuitive control theory lecture note, designed for engineering applications, in both Chinese and English. — control theory notes.

AnyMath - AI 动画生成器 | 在线创建数学动画 — an online math animation generator, useful for quick teaching demos.

Wing900/PlotKityCat — a desktop app for mathematical teaching visualization, built with Wails + Vue + Go + Python Runtime.

Theorems of the 21st century — records and introduces important mathematical theorems from the 21st century.

机器人工程师学习计划 - 知乎 — learning roadmap for robotics engineers.

机器人工程师本科学习计划（试读版） - 知乎 — undergraduate robotics study plan, trial version.

机器人工程师本科学习计划（试读版） - 飞书云文档 — Feishu document version of the robotics study plan above.

Exorust/TorchLeet: Leetcode for Pytorch — PyTorch exercises in a LeetCode style for practicing tensor operations.

Polypad – The Mathematical Playground — an interactive mathematical whiteboard for geometry, algebra, and classroom demonstrations.

PyTorch vs TensorFlow Syntax: 15 Operations Side-by-Side – TildAlice — side-by-side syntax comparison of common PyTorch and TensorFlow operations.

retikz — docs for a web-based TikZ/flowchart drawing tool.

GitHub: Pionpill/retikz: 使用 react，d3 实现 tikz 绘制流程图 — source repository for retikz.

guchuanv-alt/SimART — a realistic multimodal simulation platform for 6G ISAC, combining robot simulation, ray tracing, and wireless evaluation.

WLJS Notebook — an open-source notebook interface based on Wolfram Engine.

WLJSTeam/wljs-notebook: Open-source notebook interface for Wolfram Engine — source repository for WLJS Notebook.

Erdős planar unit-distance — discussion record on the Erdős planar unit-distance problem.

【【解线性方程组】的现代姿势：为什么不应该求逆矩阵？】 — explains why numerical linear algebra usually should not explicitly invert matrices.

終 · 一步到位的张量指标运算 - 知乎 — a long article on tensor index notation and tensor operations.

万字全景：超声波雷达和毫米波雷达的强劲对手——UWB近场感知技术 - 知乎 — introduction to UWB radar.

RCFS — Robotics Codes From Scratch, an example toolbox for robot optimal control and trajectory optimization.

Academic Platforms / Papers / Blogs

MLOps Community | Learn, Meet & Grow in Real-World MLOps — an MLOps community with articles, events, and real engineering experience.

Discover AI/ML SOTA papers with code and benchmarks - Wizwand — discovery platform for AI/ML papers, code, and benchmarks.

Papers with Code — aggregator for papers, code, and task leaderboards.

Azimuth — John Baez’s blog on math, physics, and complex systems.

Distill — Latest articles about machine learning — visual-explanation articles on machine learning; good for collecting classics.

Awesome Architecture — architecture learning resource navigation.

Gregory Gundersen — a blog with lots of statistics content.

A History of Large Language Models — Gregory Gundersen’s article on the history of large language models.

Design / Creation / Web Tools

SiriusFzh/NovaForge: 锻造你的知识体系 | 考研·考公·专业课·科研·项目·竞赛通用笔记模板。LaTeX+Markdown双版本，7步模块化结构，5种场景模板，一键换色，开箱即用。 — a general knowledge-system and note template for exams, research, projects, and competitions.

ianho7/maptoposter-online: ✨❤️‍🔥Turn the cities you love into stunning designs — online design tool for turning city maps into posters.

be5invis/Iosevka: Versatile typeface for code, from code — a customizable monospace typeface for coding and mathematical typography.

LaTeX Theme Editor — online editor for LaTeX themes and color palettes.

kocurvik/beampptx: Convert LaTeX Beamer slides to PowerPoint with flawless vector graphics and embedded video support. — converts LaTeX Beamer slides to PowerPoint with vector graphics and embedded video support.

Image Hosting / Media Hosting

A collection of image and media hosting services, useful for temporary sharing or external-link fallbacks.

• SM.MS https://sm.ms/: common free image hosting service.
• Streamja https://streamja.com/: video hosting and sharing service.
• 路过图床 https://imgtu.com/: Chinese image upload and external-link service.
• 聚合图床 http://tool.mkblog.cn/tuchuang/: multi-image-host upload aggregator.
• 聚合图床-免费无限图片上传 https://www.superbed.cn/: aggregated image-hosting service.
• Upload.cc 免費圖片上傳網 https://upload.cc/: free image upload and sharing service.
• Postimages https://postimages.org/: commonly used image upload and forum-link service.
• 多图片上传 https://yytu.xyz/: multi-image upload service.
• 云转码 https://v.superbed.cn/: media transcoding and hosting service.
• 笑果图床 https://imagelol.com/: image upload and external-link service.
• 小鹿快传 https://deershare.com/send: file transfer service.
• LskyPro https://www.lsky.pro/: open-source image-hosting program for self-hosting.
• 即刻图床 https://jiketuchuang.com/: Chinese image upload and external-link service.
• 图壳 https://imgkr.com/: image upload and hosting service.
• 老王图床 https://www.gejiba.com/: image upload and external-link service.
• Mikupics https://mikupics.cn/: image hosting service.
• Imgur The magic of the Internet https://imgur.com/: widely used international image hosting and sharing community.
• Streamvi Video Sharing Made Simple https://streamvi.com/: video upload and sharing service.
• 去不图床 https://7bu.top/: image upload and external-link service.
• 图床导航 http://imgdh.com/: image-hosting service navigation.
• Z4A图床 https://www.z4a.net/: image upload and external-link service.
• ImageHub https://www.imagehub.cc/: image hosting service.
• IMG. 全球CDN免费图床 https://img.: likely incomplete link; can be cleaned later.
• 老王图床 https://img.gejiba.com/: upload entrance for 老王图床.
• 图林 https://imglink.org/: image-link and hosting service.
• ImageOss-免费图片存储空间 https://www.imageoss.com/: image storage and external-link service.
• 图片存储 https://pic.xywm.ltd/: image upload and storage service.
• 幻域图床 https://tc.rpgsky.net/: image upload and external-link service.
• 跬步图床 https://img.kuibu.net/: image upload and external-link service.
• QQ.TC http://qq.tc/: short-link or image-link service entrance.
• 小波图床 https://www.tuchuangs.com/: image upload and external-link service.
• 麦克图床 https://macimg.com/: image upload and external-link service.
• imgsha https://imgsha.com/: image hosting service.
• ImgBB https://imgbb.com/: common international image upload and hosting service.
• StreamSB-VideoPlatformForYou https://streamsb.com/: video hosting platform.
• 最图床 https://z.photos/: image upload and external-link service.
• MBA图床 https://pic.mba/: image upload and external-link service.
• Imagecache & resize service https://images.weserv.nl/: image caching, proxying, and resizing service.

Fonts / Typography Configuration

Use separate VS Code font settings for Markdown / LaTeX / TeX so they use serif fonts that are more comfortable for reading.

{
  "editor.fontFamily": "'JetBrains Mono', 'SF Mono', Menlo, monospace",
  "editor.fontSize": 14,
  "editor.lineHeight": 1.45,

  "[markdown]": {
    "editor.fontFamily": "'New York', 'Songti SC', 'LXGW WenKai Screen', serif",
    "editor.fontSize": 16,
    "editor.lineHeight": 1.65,
    "editor.wordWrap": "bounded",
    "editor.wordWrapColumn": 88
  },

  "[latex]": {
    "editor.fontFamily": "'New York', 'Songti SC', 'LXGW WenKai Screen', serif",
    "editor.fontSize": 16,
    "editor.lineHeight": 1.65,
    "editor.wordWrap": "bounded",
    "editor.wordWrapColumn": 88
  },

  "[tex]": {
    "editor.fontFamily": "'New York', 'Songti SC', 'LXGW WenKai Screen', serif",
    "editor.fontSize": 16,
    "editor.lineHeight": 1.65,
    "editor.wordWrap": "bounded",
    "editor.wordWrapColumn": 88
  }
}

hydropix/TranslateBooksWithLLMs: Translate full-length books and documents with Ollama, OpenAI (comptatible), Gemini, Mistral, Poe or OpenRouter. Preserves formatting. Resumes where you left off. No file size limits. — translate full books.

rbc33/Ghostbar: Free macOS-native Cluely alternative. AI chat invisible to screen sharing. Ollama, OpenAI, Claude. Swift. No Electron. No subscription. — a tool that hides AI chat windows during screen sharing.

AI coding desktop pets made by yetone:

• Website: Confirmo
• Gallery: Confirmo Sprites - Share Custom Characters

Codex pet galleries:

• Codex Pets — Codex Pets
• Petdex — 适用于 Codex 的动画伙伴

OpenPets - desktop pets for coding agents

huggingface/nanowhale — nano deepseek-v4.

A reinforcement-learning tutorial:

• GitHub: walkinglabs/hands-on-modern-rl: 🚀 An open-source, hands-on curriculum bridging the gap from basic RL concepts to LLM alignment, RLVR, and advanced Agentic systems.
• Ebook: https://walkinglabs.github.io/hands-on-modern-rl/

Xiaomi open-sourced OmniVoice, a voice-cloning TTS model that supports 600+ languages.

• WeChat article: 超600种语言，一个模型全搞定! 小米开源 OmniVoice 多语言语音克隆 TTS
• GitHub: k2-fsa/OmniVoice: High-Quality Voice Cloning TTS for 600+ Languages

陪读蛙 — an open-source alternative to Immersive Translate.

Coding/Token Plan 简览（付费） - 文档共建 - LINUX DO — Coding Plan roundup.

GitHub - liliMozi/openhanako: A personal AI agent with memory, personality, and autonomy — built on Electron. · GitHub — an agent with a pleasant color palette, based on Pi and built with Electron.

别让 AI 工具每次重新认识你 — neuDrive GitHub: agi-bar/neuDrive

esengine/DeepSeek-Reasonix: DeepSeek-native AI coding agent for your terminal. Engineered around prefix-cache stability — leave it running. — a terminal AI coding assistant built specifically for DeepSeek, with a focus on prefix-cache stability.

最近使用 ChatGPT 5.5 Pro 的体验 | Gowers 的博客 — A recent experience with ChatGPT 5.5 Pro | Gowers’s Weblog

QingJ01/DeepSeekCode: 基于 Claude Code 代码库改造的本地 CLI 编程代理

ml1301215/vibe-proving-math: Math reasoning & proof-assist stack: GVR loop, citation checks via TheoremSearch, streaming API + web UI.

macOS / Productivity Tools

Sanyam-G/switch: A keyboard-driven window switcher for macOS. Notarized. Free. — a macOS window switcher.

duongductrong/Snapzy: An open-source native macOS screenshot and screen recording app. A CleanShot X alternative.

marsmay/UsageBoard: macOS menu bar app for aggregating API usage quotas via plugin scripts — a new code-usage monitoring tool, implemented as a macOS menu bar app with plugin scripts.

AndDrive - Android in Finder, Mirroring, and Clipboard Sync on Mac

Research / Math / Optimization

guofei9987/scikit-opt: Genetic Algorithm, Particle Swarm Optimization, Simulated Annealing, Ant Colony Optimization Algorithm,Immune Algorithm, Artificial Fish Swarm Algorithm, Differential Evolution and TSP(Traveling salesman)

MATLAB & SIMULINK AI Toolkits:

• matlab/matlab-agentic-toolkit: The MATLAB Agentic Toolkit brings trusted MATLAB capabilities to AI agents, making engineering and scientific workflows agent-ready.
• matlab/simulink-agentic-toolkit: The Simulink Agentic Toolkit gives your AI agent both the tools and the expertise to work effectively with Simulink and Model-Based Design.

张量网络注意力 — Tensor Network Attention — also introduces tensor-network operations. Related material appears in the tensorbook inside thomasahle/tensorgrad: Machine Learning with Symbolic Tensors, whose Python library can do symbolic tensor computation. This paper gives a concrete introduction: arxiv.org/pdf/1603.03039. There is also a visualization tool: Tensor Graph Visualizer

Learning Beyond Gradients — the author proposes Heuristic Learning.

Ryan Tibshirani: Teaching — courses on optimization, statistical learning, and data science.

【硬核科普】WiFi是怎么传递信息的？把信息装进电磁波有多难？ - YouTube — feels quite good, though the voiceover is hard to listen to.

【漫士】你看这个世界的每一眼，都在进行着深刻的运算 - YouTube — an excellent video.

有哪些算法惊艳到了你？ - 北冥有鱼的回答 - 知乎 — introduces Newton’s method.

Think Linear Algebra — Think Linear Algebra

ChenLiu-1996/figures4papers: My Python scripts to make high-quality figures for publications in top AI conferences and journals. — figure scripts shared by a PhD student.

Aurora：面向矩形矩阵的杠杆感知优化器 | Tilde — Aurora: A Leverage-Aware Optimizer for Rectangular Matrices | Tilde — a new optimizer, Aurora, addressing the issue where Muon can cause some neurons or parameters to go dead.

Academic Resources / Literature Platforms

SciPost Homepage

anonymous github — an anonymous open-source platform for anonymous submissions.

高校名录数据可视化 — could be useful for college application decisions.

Web / Development / Creation Tools

1001 online tools; I do not think I have seen a larger collection.

• Website: WebUtils - 纯前端工具集
• GitHub: chicogong/html-tools: 🛠️ 1001+ 纯前端在线工具集 | 单文件·零构建·可离线·隐私优先 | Pure HTML Tools - Single file, Zero build, Offline-ready, Privacy-first

one-ea/Monolith: ⚡ A premium serverless edge-native blog — Cloudflare Workers + Hono + React + pluggable storage adapters — a blog framework.

DimensionDev/Flare: Browse Mastodon, Bluesky, X, Misskey, Nostr and RSS all in one app. One timeline, all your accounts, cross-post everywhere. — a social-media aggregator supporting Mastodon, Bluesky, X, Misskey, Nostr, and RSS.

Cicada — a web whiteboard tool.

How LLMs Work — A Visual Deep Dive

The fgruler package can draw rulers in the page foreground. The example uses:

\usepackage[type={alledges,alledgesT}, color=blue]{fgruler}

Here alledges means drawing rulers on all four page edges, alledgesT means the four edges use the “T” style, and color=blue sets the ruler color to blue.

Music Download

• CharlesPikachu/musicdl: Musicdl: A lightweight music downloader written in pure python. (轻量级无损音乐下载器，支持数十个音乐/有声读物平台，例如网易云音乐，QQ音乐，酷狗音乐，酷我音乐，咪咕音乐，千千静听，汽水音乐，Bilibili，街声，喜马拉雅，懒人听书，荔枝FM，蜻蜓FM，JOOX，TIDAL，YouTube，Apple Music，Spotify，Qobuz，SoundCloud等主流音乐平台)
• MrsEWE44/musicDownload

GitHub - datalab-to/chandra: OCR model that handles complex tables, forms, handwriting with full layout. · GitHub — an OCR model that automatically handles layout.

GitHub - uutils/coreutils: Cross-platform Rust rewrite of the GNU coreutils · GitHub — GNU Coreutils rewritten in Rust.

A control-theory algorithm simulation tool: Control Theory Simulator GitHub - cool-koala/Control-Theory-Simulator · GitHub

魔方维基和谜题 - Ruwix.cn — cube wiki and puzzles.

GitHub - mmulet/term.everything: Run any GUI app in the terminal❗ · GitHub

Scientific & Scholarly Research Database | Scilit — a literature data center.

AI & Tech Blog | Guides, Reviews & Insights | Dupple — an AI-focused blog.

HeyBlog Console — a webpage that crawls friend links and aggregates blogs.

MathArena — a benchmark leaderboard for large-model math ability.

GitHub - atuinsh/atuin: ✨ Making your shell magical · GitHub — enhanced shell-history search.

【漫士】“我证明了黎曼猜想，AI说我是对的” - YouTube

1. The video team built a test set roughly comparable to college-entrance-exam difficulty, with trap questions that can mislead AI models. Overseas models were led by Gemini, GPT, and Claude; domestic models by Seed and DeepSeek.
2. Current AI math ability is in a contradictory state. It is good at brute-force exploration and can use powerful computation plus existing knowledge to solve complex calculation and case-analysis problems. But for math that requires pure insight, elegant intuition, and concise construction, such as extremal constructions or combinatorics, AI still struggles. Creating new mathematical theory remains very difficult for current statistical language models.

Aibijia — AI 订阅，一键比价 — one-click price comparison for AI subscriptions.

TokenNav - AI API 中转站导航 — navigation site for AI API relay services.

Telegram Themes — create custom Telegram themes here. Some themes I made:

• Rose-Pine-Dawn for Android
• Rose-Pine-Moon for Android
• Rose-Pine-Dawn for Mac
• Rose-Pine-Moon for Mac

PaperPlotHub — Open marketplace for paper plot scripts — resources for scientific plotting.

Collected a few Git clients for Windows and Mac:

• gitextensions/gitextensions: Git Extensions is a standalone UI tool for managing git repositories. It also integrates with Windows Explorer and Microsoft Visual Studio (2015/2017/2019).
• DetachHead/rebased: A git client based on the IntelliJ platform
• GitHub Desktop | Simple collaboration from your desktop
• Sourcetree | Free Git GUI for Mac and Windows
• SourceGit - Opensource & Free Git Gui Client
• TortoiseGit – Windows Shell Interface to Git
• GitKraken Legendary Git Tools | GitKraken
• GitButler

Xiaomi MiMo 100T — Xiaomi MiMo trillion-token creator incentive program. You can apply for token access.

点点 — Xiaohongshu’s AI Chat.

Dokobot - AI Agent 的浏览工具 — a CLI tool for getting information from the browser.

GitHub - rossant/awesome-math: A curated list of awesome mathematics resources · GitHub — exactly what the name says: Awesome Math.

Linear Algebra Cheat Sheet by gustavhelms - Download free from Cheatography - Cheatography.com: Cheat Sheets For Every Occasion — a linear algebra cheat sheet. The site has many other cheat sheets too.

Introducing talkie: a 13B vintage language model from 1930 — a 13B language model trained on pre-1930 material. Very back-to-basics.

GitHub - warpdotdev/warp: Warp is an agentic development environment, born out of the terminal. · GitHub — Warp is now open source.

Introducing Laguna XS.2 and Laguna M.1 — Poolside — two new models.

GitHub - caojiaolong/spaces-index: 本项目自动抓取并索引科学空间的文章元数据，按研究主题进行规则分类，方便在 GitHub 上快速浏览并跳转到原文。 · GitHub — an index of Su Jianlin’s Scientific Spaces articles, automatically crawled and categorized by research topic.

GitHub - Hmbown/DeepSeek-TUI: Coding agent for DeepSeek models that runs in your terminal · GitHub — a deepseek-tui developed by an American developer.

Command Code - AI coding agent with taste — offers a $1 plan.

During the May Day holiday I had headaches for five or six straight days. Looking at the left side of the screen made my eye uncomfortable, with a pulsing pain behind the eye and a pulling sensation. I suspect muscle fatigue from too much screen time and staying up late. I spent almost the whole holiday lying in bed sleeping and did almost nothing. I really need to pay attention to health, eye care, and rest. It is awful when everything feels bad and sleep is the only option; soon after waking up each day, the headache would start again. From now on I want to gradually build better routines: sleep and wake on time, use my eyes scientifically, reduce screen time, go outdoors more, and eat healthier.

This was a busy week: deepseek-v4, gpt-5.5, qwen3.6-max, kimi-k2.6, MiMo-V2.5, and HY-3.0 all launched one after another.

deepseek_v4_rolepaly_instruct/README.md at main · victorchen96/deepseek_v4_rolepaly_instruct · GitHub — a DeepSeek v4 roleplay guide.

ZCode - Simple, Fast, Vibe‑Ready — Zhipu’s zcode switched from Tauri to Electron, which is a little funny.

飞搜 - 飞书文档搜索引擎 | AI智能搜索飞书知识库、教程与干货资源 — a Feishu document search engine, using AI to search Feishu knowledge bases, tutorials, and resources.

AI Coding / Agent Platforms

pi-gui — A native desktop for AI coding agents — Pi’s GUI, somewhat like Codex App.

GitHub - multica-ai/multica: The open-source managed agents platform. Turn coding agents into real teammates — assign tasks, track progress, compound skills. · GitHub — an open-source managed agents platform that turns coding agents into teammates.

GitHub - golutra/golutra · GitHub — another agent coding platform.

Slock - Where humans and AI agents collaborate — a collaboration platform for humans and AI agents.

Kollab — a multi-agent collaborative coding platform.

GitHub - mindfold-ai/Trellis: The best agent harness. · GitHub — an agent orchestration framework.

Mercury — Soul-Driven AI Agent with Permission Guardrails | Cosmic Stack — a new 24/7 agent.

Introducing Parallel Agents in Zed — Zed’s Blog — Zed now supports parallel multi-agent views, improving AI-assisted coding workflows.

AI Tools

Introducing Arrow 1.1, a step forward in SVG generation – QuiverAI — an SVG generation model. You can try it on the official site. Vector conversion example: Drawing an app icon:

GitHub - mnfst/manifest: Smart Model Routing for Personal AI Agents. Cut Costs up to 70% 🦞👧🦚 · GitHub — model routing that automatically decides which model to use for each task to save cost.

Euphony: Visualize Chat Data in Your Browser — an official Codex community plugin for viewing conversation data.

Research & Papers

Literature | SigSep — a survey of music separation, with a corresponding open-source GitHub repository.

\[2604.10780\] LIDARLearn: A Unified Deep Learning Library for 3D Point Cloud Classification, Segmentation, and Self-Supervised Representation Learning — LIDARLearn, a unified deep learning library for point-cloud classification, segmentation, and self-supervised representation learning. GitHub - said-ohamouddou/LIDARLearn: LIDARLearn: A Unified Deep Learning Library for 3D Point Cloud Classification, Segmentation, and Self-Supervised Representation Learning · GitHub

13+ Attention Mechanisms You Should Know — a roundup of papers behind 13 types of attention mechanisms.

MathNet: A Global Multimodal Benchmark for Mathematical Reasoning and Retrieval — MIT’s global multimodal benchmark for mathematical reasoning and retrieval.

GitHub - NawfalMotii79/PLFM_RADAR: Open-source, low-cost 10.5 GHz PLFM phased array RADAR system · GitHub — an open-source, low-cost 10.5 GHz linear-frequency-modulated phased-array radar system with hardware and FPGA implementation. The processing flow is basic FFT plus CFAR detection.

Academic & Learning Resources

LeanSearch-Find theorems in Mathlib4 using natural language query — a Lean4 theorem search engine that supports natural-language queries over Mathlib4.

Matlas is a mathematical semantic search engine for searching theorems and proofs with natural language. Blog: Matlas: A Semantic Search Engine for Mathematics | FrenzyMath

互联网学术档案 — an academic paper search engine from the Internet Archive.

Type Words 官网 - 词文记 | 单词跟打 · 文章跟打 · 电脑上背单词 — a vocabulary-learning website; it feels nicely made.

https://linux.do/t/topic/1417107 — an AI-for-research post collected by a linux.do user.

ickma.dev – ∇ ickma.dev — a personal blog with course notes.

Terminal & CLI

GitHub - AnalyseDeCircuit/oxideterm: All-in-one terminal workspace — local shells, SSH, SFTP, remote IDE, AI agent, and file manager in a single native binary. Built with Tauri 2 and pure Rust SSH (no OpenSSL). Smart reconnect, MCP, RAG, plugins, 30+ themes, 11 languages. · GitHub — an SSH client implemented with Tauri.

GitHub - vercel-labs/wterm: A terminal emulator for the web · GitHub — Vercel’s open-source web terminal emulator.

MUX0 — a terminal that knows your agent — another sidebar terminal, derived from libghostty.

Developer Tools

GitHub - PerryTS/perry: A native TypeScript compiler written in Rust. Compiles TypeScript directly to executables using SWC and LLVM. · GitHub — a native TypeScript compiler written in Rust, compiling TS directly to executables.

GitHub - indaco/malt: A Homebrew client in Zig. Warm installs in milliseconds. post_install scripts that actually run. · GitHub — a Homebrew client written in Zig.

Sionna Documentation — Sionna 2.0.1 — Sionna has updated to a PyTorch version instead of the earlier TensorFlow version.

Home | D2 Documentation — a text diagram language similar to Mermaid. GitHub: https://github.com/terrastruct/d2

The Rattler Book — a Rust Conda package-manager library for building cross-platform package-management tools.

GitHub - LordAizen1/git-newspaper: Generate a Victorian broadsheet newspaper from any git repository’s history · GitHub — generate a Victorian broadsheet newspaper from a git repository’s history.

Python tools for data visualization — PyViz 0.0.1 documentation — a guide to open-source Python visualization tools.

The PyViz.org website is an open platform for helping users decide on the best open-source (OSS) Python data visualization tools for their purposes, with links, overviews, comparisons, and examples.

Design & Tools

Paper – design, share, ship — an online design and sharing platform.

GitHub - tw93/Kami: 👩‍🚒 Good content deserves good paper. · GitHub — tw93’s document design Skill.

• Web page: https://kami.tw93.fun/
• Install with: npx skills add tw93/kami

landppt — an online PPT generation tool.

editbanana — an online text-editing tool.

macOS Tools

Pica, a MacOS font management app — a nice font manager for macOS.

Browser

GitHub - caffienerd/startpage: Professional Terminal Start Page is a sleek, customizable browser start page built around an interactive terminal. It combines productivity tools like semantic AI routing, Gemini integration, real-time weather, and network utilities with a premium aesthetic and multiple themes. Designed for speed and flexibility. · GitHub — a clean terminal-style browser start page. Demo: https://caffienerd.github.io/startpage/

GitHub - dd3v/favbox: 🔖 FavBox is a local-first experimental browser extension that enhances and simplifies bookmark management without cloud storage or third-party services. · GitHub — a local-first browser extension for bookmark management.

Notes & Others

Tolaria — A second brain for the AI era — a second-brain note-taking tool for the AI era.

GitHub Next — GitHub’s official frontier projects and technology previews.

查看 IP 最近Torrent下载的内容 — view recent Torrent downloads associated with an IP address.

The AI Timeline — another AI newsletter.

AI Coding / Agent Platforms

GitHub - EKKOLearnAI/hermes-web-ui: Web dashboard for Hermes Agent — multi-platform AI chat, session management, scheduled jobs, usage analytics & channel configuration (Telegram, Discord, Slack, WhatsApp) · GitHub — Web dashboard for Hermes Agent, with multi-platform AI chat, session management, scheduled jobs, usage analytics, and channel configuration for Telegram, Discord, Slack, and WhatsApp.

GitHub - CodebuffAI/codebuff: Generate code from the terminal! · GitHub — an open-source code agent. Free version: Freebuff — the free coding agent

https://github.com/zhukunpenglinyutong/desktop-cc-gui — a Claude Code GUI client and vibe-coding platform for developers.

GitHub - getpaseo/paseo: Orchestrate coding agents remotely from your phone, desktop and CLI · GitHub — similar to the earlier lody.ai.

Collections of multi-agent aggregation platforms: Depo | AI Harness tools Lex’ Notes of Multi-Agent Coding tools There are a lot of these tools lately. I have tried a few myself and most feel somewhat awkward; I still prefer the official CLI, Codex App, or maybe Warp. Warp Terminal now supports vertical tabs, notifications, and remote control. Pretty nice.

AI Usage Monitoring

Tools for monitoring AI coding plan usage. For now I recommend the third one, OpenUsage.

1. GitHub - tddworks/ClaudeBar: A macOS menu bar application that monitors AI coding assistant usage quotas. Keep track of your Claude, Codex, Antigravity ,and Gemini usage at a glance. · GitHub
2. GitHub - steipete/CodexBar: Show usage stats for OpenAI Codex and Claude Code, without having to login. · GitHub
3. GitHub - robinebers/openusage: Burning through your subscriptions too fast? Paying for stuff you never use? Stop guessing. OpenUsage is free and open source. · GitHub
4. GitHub - Four-JJJJ/AI-Plan-Monitoring: macOS menu bar app to monitor AI plan limits, relay balances, and local Codex accounts · GitHub
5. GitHub - nguyenphutrong/quotio: Stop juggling AI accounts. Quotio is a beautiful native macOS menu bar app that unifies your Claude, Gemini, OpenAI, Qwen, and Antigravity subscriptions – with real-time quota tracking and smart auto-failover for AI coding tools like Claude Code, OpenCode, and Droid. · GitHub

Codex Usage

GitHub - Dailin521/codex-provider-sync: Synchronize Codex session provider metadata across rollout files and SQLite state. · GitHub — if you use both official subscriptions and third-party APIs, Codex history may fail to display because the provider identifiers differ. This tool synchronizes them so the history shows up correctly.

If you use the official subscription, you may need TUN mode to avoid reconnecting issues. If you do not want to enable TUN mode, create a .env file under ~/.codex and add proxy settings like:

HTTP_PROXY="http://127.0.0.1:your-proxy-port" # clash seems to use 7897 by default
HTTPS_PROXY="http://127.0.0.1:your-proxy-port"
NO_PROXY="localhost,127.0.0.1"

Agent Skills

GitHub - labarba/sciwrite: Agent Skill for AI-assisted manuscript writing review, based on Dr. Kristin Sainani’s “Writing in the Sciences” methodology. · GitHub — a scientific writing review Skill based on Kristin Sainani’s Writing in the Sciences methodology.

GitHub - lewislulu/html-ppt-skill: HTML PPT Studio — AgentSkill with 24 themes, 31 layouts, 20+ animations for building professional HTML presentations · GitHub — HTML PPT Studio Skill, with 24 themes, 31 layouts, and 20+ animations.

GitHub - ZreadAI/zread-skill: Zread Skill · GitHub — Zread released a Skill that lets AI call the local repository wiki generated by Zread. It may not always be necessary, but it decouples the agent from reading the repository directly.

Research & Papers

\[2604.12968\] Evolution of Optimization Methods: Algorithms, Scenarios, and Evaluations — a survey of optimization methods, covering algorithms, scenarios, and evaluation methods.

\[2604.01472\] The Newton-Muon Optimizer — Muon is essentially an implicit Newton method.

GitHub - SensRad/vegvisir: A Combined SLAM and Localization System for both RADAR and LiDAR · GitHub — a SLAM and localization system that supports both RADAR and LiDAR.

Academic & Learning Resources

GitHub - JabRef/jabref: Graphical Java application for managing BibTeX and BibLaTeX (.bib) databases · GitHub — an open-source reference manager for BibTeX and BibLaTeX databases.

The world’s largest collection of open access research papers — CORE, the world’s largest collection of open-access research papers.

Stanford CS336 — the 2026 version of CS336.

Raspberry Pi 树莓派 (官网26年04月更新)

LibreCommons — a textbook resource library with free and open textbooks, courses, and learning materials across many disciplines.

Terminal & CLI

GitHub - muxy-app/muxy: A macOS terminal multiplexer built with SwiftUI and libghostty · GitHub — another vertical-tab terminal, similar to cmux and related tools.

GitHub - thegdsks/awesome-modern-cli: A curated list of modern alternatives to classic command-line tools. Faster, prettier, smarter replacements for the Unix utilities you use every day. · GitHub — a curated list of modern CLI alternatives to classic Unix tools.

Numbat - scientific calculator — a calculator with units.

Developer Tools

GitHub - indragiek/GHFS: Mount GitHub repositories as a virtual read-only macOS filesystem · GitHub — mount GitHub repositories as a virtual read-only filesystem on macOS.

Mounting an SSH server filesystem locally can make editing and managing remote files more efficient: GitHub - libfuse/sshfs: A network filesystem client to connect to SSH servers · GitHub https://github.com/winfsp/sshfs-win

GitHub加速下载代理 - 快速访问 GitHub 文件 — a GitHub mirror/proxy, with Docker and SourceForge acceleration plus a developer toolbox: https://tool.gh-proxy.com/ or https://ittool.wort.uk/

GitHub - lra/mackup: Backup and keep your application settings in sync. · GitHub — synchronize and back up application settings; handy when switching machines.

chezmoi - chezmoi — manage dotfiles across machines.

GitHub - floccusaddon/floccus: :cloud: Sync your bookmarks privately across browsers and devices · GitHub — private bookmark sync across browsers and devices, supporting Firefox, Chrome, and mobile apps.

dmcheck - 域名查询

Notes & Blogs

GitHub - Shpigford/clearly: Markdown editor and knowledge base for Mac · GitHub — a pure Swift markdown editor for macOS: https://clearly.md/

Bear Blog — a minimalist blogging platform with no JS, no tracking, and no ads.

macOS Tools

Mac Menubar Apps - Curated list of the best menubar apps for your life

The Google app for desktop — Google’s Windows app.

Design & Themes

Premium Interactive Web Sections — high-quality interactive web components for modern web design.

Something beautiful, Rose Pine theme — Rose Pine color themes, with many community ports for apps and tools. Similar themes:

• Dracula Theme for 400+ apps
• Tokyo Night Website
• Everforest
• Nord
• Catppuccin

GitHub - pbakaus/impeccable: The design language that makes your AI harness better at design. · GitHub — an AI design tool that provides a design language to help AI tools design better. The author clearly has taste; the website is elegant: https://impeccable.style/

GitHub - KurtGokhan/tegaki: Handwriting animation for the web. Supports any font or text. · GitHub — handwriting animation for the web, supporting any font or text.

Fonts & Typography

What’s Trending in Type · Typewolf — tracks trends in web typography.

Fonts In Use - Fonts In Use — real-world typography examples.

Butterick’s Practical Typography — Butterick’s practical typography guide.

Font.icu - 字体预览与下载

Images & Screen Recording

长图拼接工具在线使用_垂直水平无缝拼接_聊天记录截图长图 — online long-image stitching tool for seamless vertical or horizontal stitching, useful for chat screenshots.

GitHub - blueberrycongee/CursorLens: Create stunning screen recordings for free. Open-source, no subscriptions, no watermarks, and free for commercial use. An alternative to Screen Studio. · GitHub — another open-source screen recording tool. I had shared Recordly before, and there also seems to be Open Screen.

Resource Communities

SeedHub | 影视&动漫分享 — film and anime resource sharing site.

Depo | A curated collection of resources — a developer-oriented resource collection site.

Cuflow – AI-Powered Study & Productivity Tool — AI-powered study and productivity tool.

The skillful toil and the wise worry, while the incapable seek nothing;
Fed and roaming freely, like an unmoored boat.
Why bother with worldly clouds? Better to rest and feast.

moss — Markdown Workspace.

GitHub - gunnargray-dev/unicode-animations — Unicode spinner animations as raw frame data. npx unicode-animations --web

GitHub - jgraph/drawio-mcp — draw.io MCP, providing multiple ways to connect with AI, e.g., directly: https://mcp.draw.io/mcp

Coding Plan Benchmark — Domestic Coding Plan evaluation. Coding Plan Comparison — Chinese AI Coding Plan comparison (GLM/MiniMax/Kimi/qwen, etc.).

GitHub - ztx888/HaloWebUI — OpenWebUI fork with Chinese interface for better Chinese user experience, added model billing and usage statistics.

GitHub - scipenai/tylax — Bi-directional converter between Typst and LaTeX. Available as CLI tool and Web interface.

GitHub - Jermic/awesome-aiart-pics-prompts — Web: https://aiart.pics/ — 3000+ Gemini Nano Banana Pro high-quality prompts and generated examples covering photography, design, art, marketing, etc. Bilingual support, JSON format.

Multiple Solutions to a Junior High Math Extremum Problem — Detailed explanation of Cauchy-Schwarz inequality, geometric method, function method for high school math maximum problems. Author saw this repeatedly on short videos and wrote this article.

Murfy, Free Online LaTeX Editor — AI LaTeX online editor, free tier is decent, better than Overleaf.

GitHub - stablyai/orca — An agent IDE.

GitHub - opendatalab/MinerU-Document-Explorer — MinerU’s knowledge base document Wiki tool with MCP support.

zhangmeishan.github.io/teaching.html — Large model course from a HIT professor.

GitHub - gsd-build/get-shit-done — Lightweight meta-prompting, context engineering and spec-driven development system for Claude Code by TÂCHES. Also applicable to agents other than Claude Code.

iFixit: Free Repair Manuals — Free repair manuals, pretty interesting.

BuildCores: 3D PC Builder and Part Picker — Online PC building simulator, fun.

Hermes Agent — Agent similar to OpenClaw. hermes-agent/skills/creative/manim-video — Hermes launched manim Skill, letting AI create demonstration videos. Manim is the tool used by 3Blue1Brown for video production.

GitHub - Ataraxy-Labs/opensessions — tmux sidebar with agent status, git branches, and instant switching.

GitHub - alibaba/Logics-Parsing — Alibaba’s new OCR model.

Open-source screen recording tools:

• GitHub - siddharthvaddem/openscreen — Create stunning demos for free. Open-source, no subscriptions, no watermarks, free for commercial use. Alternative to Screen Studio.
• Recordly — Open-source app for incredible screen recordings.

GitHub - Wangnov/uv-custom — Chinese mirror for uv and uv python runtime for high-speed downloads.

\[2604.02176\] Adam’s Law: Textual Frequency Law on Large Language Models — LLMs should pay more attention to higher-frequency expressions with the same meaning.

\[2604.04771\] MinerU2.5-Pro — MinerU’s new model pushing the limits of data-centric document parsing at scale.

David Noel Ng — Blog focusing on LLM mechanisms, AI safety & alignment, biophysics & hardware engineering. Summary: https://chatglm.cn/share/PF1kUqbD

GitHub - microsoft/VibeVoice — Microsoft’s open-source speech model, including TTS and ASR.

预训练经验 — Pre-training experience (Zhihu).

GitHub - whwangovo/pyre-code — Self-hosted ML coding practice platform. 68 problems from ReLU to flow matching — attention, training, RLHF, diffusion, and more. Similar to TorchCode mentioned earlier. Instant feedback in browser.

Helium Browser Update — Supports more compact appearance, vertical tabs, zen mode, Safari-like themes. Ready to fully switch to Helium, this is the browser I love.

Superconductor — Multi-Agent aggregation platform built purely in Rust.

GitHub - MiniMax-AI/cli — MiniMax CLI.

In-Place Test-Time Training — Automatically adjust some weights during testing. Neural Computers — Neural computers, quite interesting.

AlgoMaster Newsletter — Algorithm-focused newsletter. For example, LeetCode was HARD until I Learned these 15 Patterns summarizes 15 common LeetCode patterns: prefix sum, two pointers, sliding window, fast-slow pointers, etc.

git-star — GitHub repository ranking by star count, with various language and category rankings to see which projects are popular.

GitHub - casey/just — Command runner. Initially I wanted a Make alternative, asked AI and discovered this popular tool.

MiniMax Audio: Make Inspiration Sound Pleasant — MiniMax’s new music generation model.

Competition Discovery — Data science & AI competition aggregation platform. Subscribe to email for new competition notifications.

GitHub - hitmux/HitCC — Complete reverse-engineering documentation of Claude Code CLI Node.js version v2.1.84.

GitHub - ayaka-notes/overleaf-pro — Online LaTeX editor built on open-source Overleaf, supporting many closed-source features like Git integration.

GitHub - Yousa-Mirage/Tufted-Blog-Template — Tufte-style blog website template powered by Typst. Use Typst to create your personal website.

GitHub - iOfficeAI/OfficeCLI — Office CLI, the world’s first and best Office suite for AI agents. Free, open-source command-line tool for AI agents to read, edit, and automate Word, Excel, and PowerPoint files. Single binary, no Office installation required.

Zhipu and Qwen’s international sites are cleaner than domestic versions, currently not yet paid. Z.ai - Free AI Chatbot & Agent — Zhipu GLM international site with chat and agent modes. Agent mode can integrate into IM, currently supporting Feishu, WeChat, Telegram, and Discord. Qwen Chat — Alibaba Qwen international site.

• LaTeX rendering plugin for terminal, suitable for Obsidian usage. GitHub - MaxwellsEquation/LaTerM-obsidian GitHub - MaxwellsEquation/LaTerM

GitHub - e10nMa2k/cc-mini — Another Claude Code implementation.

Metis — Curator Dashboard — Automatically discover new projects from GitHub trending and Product Hunt.

GitHub - zai-org/GLM-skills — Zhipu released GLM5v model along with open-sourced Skills.

GitHub - onyx-dot-app/onyx — Open Source AI Platform for AI Chat with advanced features, works with every LLM.

GitHub - AnalyseDeCircuit/oxideterm — All-in-one terminal workspace: local shells, SSH, SFTP, remote IDE, AI agent, and file manager in a single native binary. Built with Tauri 2 and pure Rust SSH (no OpenSSL). Smart reconnect, MCP, RAG, plugins, 30+ themes, 11 languages.

kexue.fm/archives/11693 — Brief introduction to median, describing how the mean is actually $\ell_{2}$ norm minimization while the median is $\ell_{1}$ norm minimization.

浅谈一下 World Model — About world models (sharp commentary edition).

Zread CLI — Zread CLI for analyzing local repositories using Zhipu coding subscription or your own API.

Lisette — Rust syntax, Go runtime

Slock - Where humans and AI agents collaborate

AI Coding Agents

• ccg-gateway: Gateway and configuration management for Claude Code, Codex, Gemini CLI. Ideal for scenarios where providers are unreliable and frequent switching is needed.
• Quotio: Command center for AI coding assistants.
• Agents You Don’t Know: Principles, Architecture & Engineering Practice: An engineering-focused long-form article on Agents.
• autoresearch: Automated research workflow exploration, suitable for closed-loop experimental tasks.
• Harness Engineering — Maybe the Agent Isn’t the Model’s Problem: Emphasizes the decisive role of harness design in Agent usability.
• Codex use cases: Official OpenAI Codex use case collection.
• Understand large codebases: Practical advice for understanding and modifying large repositories.
• Codex Plugins: Codex plugin system.
• Claude Code Cheat Sheet: Quick reference for common commands.

Developer Tools

• liteparse: Open-source document parser.
• type4me: Open-source macOS Chinese voice input project supporting recognition and text optimization.
• Zed text alignment update: Specific character alignment feature to improve code readability.
• Stitch: Google’s AI design tool,偏向 Figma workflow.
• storage.to: File-to-link sharing tool.
• OpenCLI Hub: CLI tool directory.
• GitWatt: Newly launched open-source project community.

Research & AI

• MinerU-Diffusion (arXiv:2603.22458): Treating document OCR as an inverse rendering problem.
• visual-attention-variants: Attention variants survey.
• ml-visuals: Machine learning paper figure templates.
• Neural-Network-Architecture-Diagrams: Neural network architecture diagram tool.

Infra & APIs

• BlazeAPI: Free AI API Gateway (1000 requests/day quota, suitable for testing).
• NetEase skills: Agent skills collection.
• entropic: Open-source project to watch.

Math & Reading

• How to express the solution to matrix equation AX+XB=C?: Covers vectorization to LU/QR/iterative methods, finally mentioning integral expressions.
• Why is range left-closed right-open?: Language design detail discussion.
• 5000-page Topology Textbook PDF: Massive reference material.

Life

• HealthTick: macOS menu bar sedentary reminder tool.
• Slay the Spire 2 Wiki: Game wiki site.

Agent Platform

• Emdash: Multi coding agent parallel collaboration platform.
• t3code: Coding agent aggregation platform.

Rendering

• RaTeX: Rust-native KaTeX compatible rendering engine.

goal-driven/readme_cn.md — An AI agent framework that achieves complex tasks through goal decomposition and execution.

CLI Tools

GitHub - jackwener/opencli — Uses playwright MCP to extract web information, turning any website into a CLI.

GitHub - LeoYeAI/openclaw-master-skills — Collection of 339+ high-quality OpenClaw skills covering AI, development, productivity, marketing, etc.

npx tokscale@latest — View local agent usage and track token consumption.

rtk — Make your AI coding agent smarter — Filter terminal output to save tokens and improve AI coding efficiency.

vPhone Virtual Phone

GitHub - 34306/vphone-aio — One-click script to run virtual phone environment. GitHub - Lakr233/vphone-cli — CLI version of virtual phone tool.

Design Tools

GitHub - open-pencil/open-pencil — AI-native design editor, open-source Figma alternative. Ardot — Tencent’s online collaborative design tool, Figma-like experience.

AI News & Information

何夕2077’s AI News — Curated AI news and tools, stay updated on AI frontier. AI资讯速览 — AI news at a glance. imjuya.github.io/juya-ai-daily/ — Daily AI news summary. linux.do AI Tech Daily — Linux.do forum’s AI tech daily.

Learning Resources

Operating System Principles (Spring 2026) — Nanjing University’s OS course (the professor is quite entertaining).

LLM Cookbook — LLM practical cookbook for systematic learning of large language models. 100 Papers with Code – Learning Note — Learning notes for 100 classic AI papers. The author has also updated 10 papers so far.

MicroGPT explained interactively — Interactive understanding of MicroGPT architecture. ScalingOpt - Optimization at Scale — Large-scale optimization research community.

LLM Architecture Gallery — Sebastian Raschka’s LLM architecture gallery with illustrated explanations of various model structures.

Privacy Search

SearXNG instances — Public SearXNG instance collection, self-hosted search engines.

Screen Recording

Recordly — Open-source screen recording software with high-quality recording support.

PDF Processing

GitHub - opendataloader-project/opendataloader-pdf — PDF parser that makes PDF data usable for AI.

AI Service Orchestration

Lody — AI CLI service aggregation platform via ACP protocol, supporting multi-device sync. Alma - Elegant AI Provider Orchestration — Elegant AI provider orchestration tool.

Subtitles & Translation

GitHub - buxuku/SmartSub — “妙幕” cross-platform subtitle tool supporting batch subtitle generation and translation (Baidu, Volcano, OpenAI, Ollama, DeepSeek, etc.). Bob — Clean and efficient translation tool. GitHub - pot-app/pot-desktop — Cross-platform hover translation and OCR software.

Key Display

Tools for displaying key operations during screen recording:

• Keyviz — Open-source free with smooth animations and high customizability
• KeyCastOW — Classic lightweight tool focused on stable and clear key display
• Kling — Lightweight without ads, clean interface, easy to use
• KeyCastr — Classic open-source choice for macOS users with customizable display
• Key Display — Chrome extension for in-browser key display
• Keylume — macOS virtual keyboard display for screen recording demos

Model Run Checking

CanIRun.ai — Check which models your machine can run. GitHub - AlexsJones/llmfit — CLI version, one command to find hardware-compatible models.

Academic Writing

Academic Phrasebank — University of Manchester’s collection of academic writing phrases. Let AI learn then write your own papers.

PPT Generation

GitHub - hugohe3/ppt-master — AI generates beautiful PPT from documents, 15 examples, 229 pages.

Skill Management

macOS only, though npx skills is probably better:

• GitHub - Shpigford/chops — Browse, edit, manage Claude Code, Cursor, Codex skills.
• GitHub - tddworks/SkillsManager — Discover and install AI coding assistant skills.

CLI to API

Wrap CLI tools into API services:

• GitHub - router-for-me/CLIProxyAPI — Wrap Gemini CLI, ChatGPT Codex, Claude Code into OpenAI/Gemini/Claude-compatible APIs.
• GitHub - lieeew/sub2api — One-stop open-source relay service, unify Claude, OpenAI, Gemini access, support shared pooling.

• arxiv-daily-researcher — Daily ArXiv paper digest推送, supporting multiple journals.
• aha — High-performance AI model inference library in Rust, supporting mainstream vision and language models.
• llm-from-scratch — CS336 course assignment implementation to understand LLM training principles.
• AutoClaw — Zhipu’s OpenClaw wrapper application, integrated into IM for AI workspace.

Academic & Research Tools

• alphaxiv-paper-lookup — Official alphaxiv Skill for快速获取 structured paper data.
• aminer-open-skill — Zhipu Aminer’s official Skill for paper and scholar lookup (includes 38.8 free credits).
• BERTopic — Transformer-based topic modeling tool with semantic understanding and automatic topic discovery.
• markdown-tex — Bidirectional Markdown ↔ LaTeX conversion library using AST as intermediate.

Claude Code Skills & Tools

• skills.sh — Vercel’s official Skill directory, supports npx skills add one-click install, recommended as first choice.
• Find awesome Agent Skills — Curated Agent Skills discovery platform.
• awesome-claude-skills — Community-curated Claude Skills list.
• SkillHub — Claude Skills marketplace.
• Tencent SkillHub — Tencent’s Skills community, optimized for Chinese users.
• claude-health — Claude Code project health check Skill by Tw92.
• CCometixLine — Rust-based Claude Code status bar tool.
• MATLAB Skills / MATLAB MCP — MATLAB’s official MCP Server and Skills collection.
• Learn Claude Code — Tutorial on building a Claude Code-like Agent from scratch, adding one mechanism at a time.
• PaddleOCR Skills — Official PaddleOCR Skill for OCR text recognition (10k pages/day quota).

Code Repository Intelligent Analysis

• Code Wiki — AI-driven code repository Wiki generator.
• DeepWiki — Conversational AI documentation for any code repository.
• Zread — AI code reading assistant that generates repository knowledge graphs.
• GitNexus — Browser-based code knowledge graph generator with Graph RAG exploration.
• Alibaba’s Qoder also has repository Wiki features.

Development Tools

• Bruno — Open-source API testing IDE, lightweight alternative to Postman/Insomnia.
• Yaak — Intuitive desktop API client supporting REST/GraphQL/WebSocket/gRPC.
• BrowserOS — Open-source Agentic browser, alternative to ChatGPT Atlas/Perplexity Comet.
• Hello Claw — Datawhale’s OpenClaw guide covering principles and implementation.

Web Content Extraction

• defuddle / Online version — Obsidian CEO’s web-to-Markdown tool, already built into official web clipper plugin. Drag to bookmark bar for quick抓取 and copy.
• markdown.new — Cloudflare-powered URL to Markdown service.
• pure.md — Global cache service between LLMs and web pages (paid).
• Jina Reader — Add r.jina.ai/ before URL to extract content.
• create360.ai — Similar to Jina Reader for web content extraction.
• agent-fetch — Web scraping tool optimized for Agents.

Utilities

• Container Image Monitoring — Docker image accelerator node status monitoring.
• 轩辕镜像 — Efficient and stable Docker image pulling service.
• Thaw — macOS menu bar management tool, successor to Ice.
• delphitools — Collection of实用小 tools: unit conversion, symbolic computation, PDF/text processing, etc.
• terraink — Map poster generator for creating unique customizable cartographic posters.
• Slidev — Web-based developer presentation tool with CLI and AI Skill support.
• snowshot / better-shot — Screenshot beautification tools.
• ekphos — Terminal-based Markdown research tool, inspired by Obsidian.

Learning Resources & Visualization

• Sorting algorithms — Simon Willison’s interactive sorting algorithm visualization.
• Principles of Sensing for Autonomy — Stanford course on autonomy system sensing principles.

Fun Projects

• WorldEnd.ai — AI-generated end-of-the-world scenarios.

macOS Tip: Skim Highlight Shortcut

Add instant highlight functionality to Skim PDF reader, simulating Sumatra PDF’s a key experience:

tell application "Skim"
    activate
    tell document 1
        set theSel to (get selection)
        set theNote to make note with data theSel with properties {type:highlight note, color:deep pink}
    end tell
end tell

Save to /Users/username/Library/Application Support/Skim/Scripts/highlight.scpt, then in System Settings > Keyboard > Keyboard Shortcuts > App Shortcuts, add Ctrl+A shortcut for Skim.

Note: When using OrbStack instead of Docker Desktop on Mac, it’s recommended to add image accelerator configuration directly in OrbStack settings rather than ~/.docker/daemon.json. Can also be configured in ~/.orbstack/config/docker.json.

WiFi DensePose → RuView

• GitHub - ruvnet/wifi-densepose

Transforms WiFi signals into real-time human pose estimation, vital signs monitoring, and presence detection — without any camera. By analyzing Channel State Information (CSI) perturbations, combining signal processing and machine learning to reconstruct human poses, breathing rate, and heartbeat.

💡 Observation: This project suddenly exploded to 15k stars. After reading an analysis on Juejin, I realized it was mainly vibe-coded. Now renamed to RuView, with a beautifully crafted README (likely all written by Claude Code). In the AI era, we need to improve our discernment and think more independently.

ACE Music

• acemusic.ai

StepFun’s music generation model.

Development Tools & Platforms

Obsidian CLI Support

Note-taking app Obsidian new version supports CLI, making it more convenient for Agent-based note management. Also supports drag-and-drop image resizing from the bottom-right corner, no longer needing third-party plugins.

TorchCode

• GitHub - duoan/TorchCode

PyTorch version of LeetCode — practice implementing softmax, attention, GPT-2, etc. from scratch with instant auto-grading. Based on Jupyter, self-hostable or use online.

Agent Skill Tool Ecosystem

Universal install: npx <tool> add <skill-repo> --all --global

AI-digest

• GitHub - jackwener/AI-digest

Automatically summarizes your Agent usage today.

Readout

• readout.org

Agent development environment dashboard.

Documentation & Academic Tools

SJTU TeX

• GitHub - sjtug/SJTUTeX

Shanghai Jiao Tong University’s LaTeX template, including dissertation, article, and presentation templates with many clever effects. Useful as reference for LaTeX template creation, or remove SJTU elements to make your own template.

AI Paper Briefing

• ai-brief.liziran.com

Daily curated paper selections with RSS subscription support.

Somark

• somark.tech

Document intelligence engine, another OCR platform like Doc2x and MinerU with good results.

Creative Tools & Visualization

Now I Get It!

• nowigetit.us

Paper2Galgame

• paper2gal.top

AI paper reader that transforms papers into visual novels.

Deep Reading & Learning Resources

Neural Scaling Laws Trilogy

• Yizhou Liu’s Blog

Trilogy on scaling laws for representations, transformations, and training.

msgsxj’s Blog Archive

• msgsxj.cn/archives

Technical blog focusing on machine learning, statistical learning, and optimization algorithms, with 33 technical posts. Several good articles, though hasn’t been updated for a few years.

Matrix Computation Courses

• Advanced Large-Scale Matrix Computation Tutorial
• Matrix Computation (ECNU)

AI & Machine Learning

Models & Frameworks

• xAI released Grok 4.2, with four Agents working simultaneously and strong search capabilities.
• Gemini 3.1 Pro: Announcing our latest Gemini AI model — Gemini 3.1 Pro is here!
• GitHub - fishaudio/fish-speech: SOTA Open Source TTS — Text-to-speech and voice cloning.
• GitHub - stepfun-ai/NextStep-1 — StepFun’s autoregressive image generation model with continuous tokens (ICLR 2026 Oral).

Papers & Research

• GitHub - MaximeVandegar/Papers-in-100-Lines-of-Code — Implementing classic AI papers in ~100 lines of code.
• Understanding Anthropic’s Interpretability Research — A nice overview of Anthropic’s interpretability work.
• On Surprising Effectiveness of Masking Updates in Adaptive Optimizers — A new optimizer: Momentum-aligned gradient masking (Magma).
• GitHub - ResearAI/AutoFigure-Edit — AutoFigure: Generating publication-ready scientific illustrations (ICLR 2026).

AI Hardware

• Taalas | The model is The Computer — Another AI inference chip company.
  • Ultra-fast HC1 chip achieving ~17k token/s inference on LLAMA 3.1 8B.
  • Try it: chat jimmy

AI Agents & Tools

• LobsterAI — NetEase’s open-source AI Agent client.
• GitHub - RightNow-AI/openfang — An Agent OS similar to OpenClaw.
• GitHub - 1rgs/nanocode — Minimal Claude Code alternative in a single Python file (~250 lines, zero dependencies).
• GitHub - minorcell/memo-code — Lightweight coding agent for terminal. Blog post discusses terminal input challenges.
• Happy - Claude Code Mobile Client — Remote-control Claude Code on mobile (though not the most user-friendly).
• GitHub - Panniantong/Agent-Reach — Let AI agents access social media and streaming platforms (Twitter, Reddit, YouTube, etc.).

Development Tools

IDEs & Editors

• GitHub - GCWing/BitFun — Agentic Development Environment with four modes: Agentic, Plan, Debug, Review.
• Markdown Now — Rust + GPUI-based markdown editor for macOS (early stage).
• {clarity} — Online editor supporting LaTeX and Typst.

Terminals

• Three terminals with vertical tab support:
  • ghostree — Worktree-Aware Terminal for Ghostty
  • cmux — Terminal built for multitasking
  • Supacode

Utilities

• GitHub - debugtheworldbot/keyStats — Keyboard & mouse usage analytics app (macOS/Windows).
• GitHub - caomengxuan666/WinuxCmd — Linux commands for Windows (900KB only).
• GitHub - xxnuo/MTranServer — Offline translation server with low resource usage.
• GitHub - tonyantony300/alt-sendme — Cross-platform file transfer without cloud storage.

Web & Browser

Web Services

• Mockdown — Free ASCII wireframe editor for quick UI prototyping.
• Scour — Semantic content aggregator mining “hidden gems” from HN, Reddit, arXiv, and thousands of blogs. Supports RSS.
• LazySo — Remove watermarks from AI-generated images (Gemini, NotebookLM, Doubao, etc.).
• markdown.new — Cloudflare-powered URL to Markdown converter.
• Wispbyte — Free 24/7 hosting.
• Reactive Resume — Free open-source resume builder.
• Trendshift — Discover trending open-source projects and developers.
• Presearch — Decentralized search engine with NSFW mode.

RSS & News

• GitHub - WCY-dt/MrRSS — Modern cross-platform AI RSS reader with Vim-like shortcuts.
• GitHub - IronsideXXVI/Hacker-News — Native macOS HN client in SwiftUI.

Research & Learning

Educational Resources

• Comprehensive AI/LLM Guide — Accessible overview of neural networks, LLM inference/training, GPU infrastructure, and Agent usage.
• CS146S: The Modern Software Developer — Stanford course on modern software development with LLM coding.
• Master the Math — Guide to rigorous mathematics of AI/ML/Data Science.

Research Tools

• ArxivLens — Search across ArXiv, PubMed, BioRxiv, MedRxiv.
• GitHub - bienz2/BibChecker — Check bibliography formatting (IEEE/ACM) without AI.
• GitHub - karakeep-app/karakeep — Self-hostable bookmark-everything app with AI tagging and full-text search.

Mathematics & Algorithms

• 3Blue1Brown Linear Algebra Notes — English and Persian notes covering key concepts.
• Elevator Scheduling Algorithms — Clear explanation of elevator/disk scheduling algorithms.
• Gaussian Mixture Models — Probabilistic model for clustering, density estimation, and anomaly detection.

Design & Media

Design Tools

• Framer China — Free Framer templates, components, and tutorials (2026).
• GitHub - lukilabs/beautiful-mermaid — Render Mermaid diagrams as SVG and ASCII art. By Craft team (who also built Craft Agent).
• GitHub - BIT-DataLab/Edit-Banana — Convert AI-generated images to editable content for PPT/draw.io.