2026年8月25日 技术热点总结

项目简介：Unofficial source-oriented reconstruction and extension of Grok Bot 0.18.0 for macOS

技术栈：TypeScript

核心介绍：This repository is an unofficial, source-oriented reconstruction of the publicly shipped Grok Bot 0.18.0 macOS app. The project began as an attempt to understand how the desktop app was put together. It now contains readable TypeScript implementations of its Electron, host, coordinator, local-execution, protocol, and renderer boundaries, plus a deterministic toolchain for turning those sources back into a working macOS It also adds a few practical experiments:

**项目数据：…

技术栈：Rust

核心介绍：walgit hosts git repositories with no database, no leader and no local state that matters. You run a single binary, point it at an S3 or GCS bucket, and you have: smart HTTP (v0/v2) fetch and push, bundle-uri clones served as static files, Git LFS, a browsing web UI, a JSON API with an SDK, per-repository push policy, webhooks — and a server that scales to repositories larger than the machine it runs on. Every machine that runs walgit is a disposable cache; the bucket is the repository.

项目数据：⭐ 1,042 Stars，🍴 56 Forks

技术栈：Python

核心介绍：لینک دونیت جهت حمایت شما عزیزان FastAPI panel for Railway + Cloudflare Workers. Manage users, inbounds, Reality / WS‑TLS configs, browser‑side IP scanning and a multi‑location Cloudflare Worker proxy — all from one dashboard. > فارسی: پنل مدیریتی و فروش سرویس VLESS با قابلیت ساخت سابسکریپشن، مدیریت کاربران و اینباندها، اسکنر آیپی (سمت مرورگر) و مدیریت Worker کلادفلر برای پروکسی چندمنطقهای.

关键特性：VLESS Reality — keys (x25519 + ML‑DSA‑65 post‑quantum) generated by the Xray binary itself, never by Python crypto. Fresh pbk / sid / spx per inbound.；**VLESS + …

项目简介：Claude Code skill for premium scroll-driven websites. Scroll becomes the timeline, on a real design floor, verified by screenshotting its own scroll.

技术栈：JavaScript、accessibility、agent-skills、ai-web-design、anthropic、claude-code、claude-code-plugin、creative-coding、design-system

核心介绍：Most AI website output fails in one of two directions. It is either well behaved and forgettable, or it is a flashy scroll animation with 2.1:1 body text, a headline that wraps to six lines on a phone, and the same six sections every other AI page has. scrollcraft is built to fail neither way: i…

项目简介：Native macOS console for herdr — all your coding agents and their live terminals, across devices

技术栈：Swift

核心介绍：The native macOS console for herdr — see Claude Code, Codex, Gemini, Grok and OpenCode across your Mac and every SSH box you own, and jump into any one of them at full-TUI fidelity. What it does · Why herdrm? · Install · Quick Start · Architecture herdr is the runtime your coding agents live on — a background server that owns their terminals and knows which one is working, blocked, or done. herdrm is a native

项目数据：⭐ 622 Stars，🍴 39 Forks

项目简介：Every coding-agent session on your Mac in one place — browse, search, resume. Rust + GPUI.

技术栈：Rust

核心介绍：A native desktop app that gathers every coding-agent session on your machine into one place — browse, full-text search, and resume any conversation in seconds. Built with Rust + GPUI (gpui 0.2 + gpui-component 0.5). macOS first; experimental Linux support since v0.2.5. Your agent history is scattered across ~/.claude, ~/.codex, and ten other private directories. Wake reads them all, read-only, and gives you one fast window into it. Everything stays local: no networ…

We present PhysCaP, a Physics-Informed Code-as-Policy agent for active perception in robotic manipulation. While vision-language-action policies excel at imitating demonstrations, they rely on passive observation and fail to infer latent physical properties critical for manipulation. PhysCaP augments code-as-policy frameworks with a physics-informed exploration layer that enables explicit information-seeking through interaction. It introduces training-free physical property extraction modules that estimate object mass and stiffness from robot proprioception without additional sensors. To ba…

Population-level behavior in large-language-model (LLM) agents cannot be characterized by single-agent benchmarks. We introduce PV-SST, a peer-voted social-platform testbed, and report a separately frozen, preregistered matched-exposure experiment spanning four topics, four unused seeds, four open-weight model families, and three prespecified larger variants. The experiment comprises 448 trials and 112 complete model-by-topic-by-seed blocks. Relative to a topic-only control, a feed of previous-round peer posts ranked by peer-generated likes increases final-round lexical similarity in both t…

We introduce Hydra-0, a generalist world model conditioned on action flow, which represents robot actions as pixel motion. This shared visual interface enables generalist world modeling and control by learning action consequences across embodiments, tasks, environments, and video-generation backbones. Our best configuration achieves 90.4% lower robot-motion error and 60.2% lower object-motion error than our action-conditioned baseline, while supporting zero-shot composition and data-efficient adaptation. On the RoboLab benchmark, Hydra-0 achieves a Pearson correlation of r=0.96 between repl…

Training-free block-sparse attention can accelerate video transformers, but row-wise attention concentration does not by itself specify an executable sparse operator. Queries sharing a block route may have poorly overlapping supports, while retained attention mass alone does not determine the post-softmax error from skipped interactions. We show that partition geometry affects both pooled support and the predictability of the remaining residual from the sparse output. We introduce SparsePR, which combines Response-Coupled Partitioning with Probe-Fitted Residual Reconstruction. Sampled-query…

Parallel reasoning improves the accuracy and robustness of large reasoning models by exploring multiple solution paths, but its computational cost grows with reasoning depth and branch count. Existing methods for managing these parallel paths typically rely on final-answer consensus, local token confidence, or isolated intermediate probes. However, these signals are often delayed, weakly tied to actual reasoning progress, or too noisy for dynamic, branch-level control. To address these limitations, we introduce ParaTempo, a training-free asynchronous parallel reasoning framework. ParaTempo …

Open-ended language-model benchmarks usually inherit a judge: a human preference panel, another model, or a brittle exact-match key. We introduce FlavourBench, an automated benchmark in which a versioned culinary system supplies dense, executable ground truth. Each task presents eight ingredients and asks for a three-ingredient portfolio; before model execution, Epicure scores all 56 possible portfolios. We evaluate 27 frontier endpoints on an identical 534-task core spanning substitution, pairing, and constrained composition. Every ranked model has exactly 89 valid responses per panel and …