2026年10月15日 技术热点总结

项目简介：Open-source iOS / macOS compatibility layer for Diablo II: Resurrected with the Reign of the Warlock expansion. Runs your own Switch copy natively on iPhone, iPad and Mac — full touch controls, 60 fps, new endgame modes. Bring your own game files. 开源的《暗黑破坏神 II：狱火重生》术士君临 iOS / macOS 兼容层

技术栈：C、compatibility-layer、diablo-2-resurrected、diablo2、game-port、ios、ipad、iphone、mac

核心介绍：This is not a tech demo that “sort of runs”. It is a mobile version polished for daily play, from your first character all the way through the endgame. Platforms: iPhone / iPad (iOS 26 and later; iOS 1…

项目简介：Let an AI model use a real iPhone with a $7 ESP32: AirPlay eyes, USB HID hands, any OpenAI-compatible vision model. No Mac, no developer account, no jailbreak.

技术栈：Python、ai-agent、airplay、automation、computer-use、esp32、esp32-s3、iphone、llm

核心介绍：English · 简体中文 Demo video and updates: X @iamhenryyuan · YouTube demo The agent sees the phone through AirPlay screen mirroring and touches it through a tiny ESP32-S3 board that plugs into the phone’s USB port as a mouse and keyboard. Any OpenAI-compatible vision model can be the brain — a cloud API, or a model you run yourself. Mo…

项目简介：A toolkit for low-level operations in DSH, designed to reduce token usage and task execution time / DSH 底层操作工具插件，旨在减少 Token 消耗与任务耗时

技术栈：JavaScript、dsh、dsh-plugin、dsh-plugins

核心介绍：Windows DSH 的性能工具插件：提供独立 bash / PowerShell 7，并用 Rust 重写文件检索与批处理， 减少反复踩 PowerShell 报错所浪费的 token、时间和模型注意力。 > tips：本项目的 ops 是广义 operations（运行操作）的缩写，代指本插件处于 dsh 和系统/代码/文件操作的基础工具位置。本项目不是 operations（运维）插件。 有问题可以加qq群：1127808570 在自用 benchmark 中获得了 的成绩，并改善了在 DSH 运行的 AI 的智力表现。 执行顺序：工具包 → bash → PowerShell 7。

项目数据：⭐ 1,545 Stars，🍴 28 Forks

项目简介：USD BRep is a standalone C++ and Python library for NURBS-based boundary representation (BRep) solid modeling

技术栈：C++

核心介绍：Standalone Omniverse repo containing USD BRep (NURBS boundary-representation solid modeling) libraries. All other build-time dependencies — OpenUSD, Python 3.12, pybind11, premake, and the repo_* tools — are pulled automatically via Packman when you build.

项目数据：⭐ 1,392 Stars，🍴 171 Forks

项目简介：Qwen-Image 2.1 NSFW/姿势出图 skill：作者级提示词配方 + NSFW LoRA + 136 姿势 VNCCS 人偶控制；适用 Claude Code / Codex / Cursor / OpenClaw

技术栈：Python、agent-skill、comfyui、image-generation、openclaw、pose-control、qwen-image

核心介绍：基于 ComfyUI + qwen_image_2.1_int8_convrot 底模（与 Civitai 作者同款底模），复刻并验证过 Civitai 作者样图（2026-10-12）。 三张图同场景递进：L0 只有提示词；L1 加 LoRA + 显式皮肤/水珠描写；L2 再加低机位与开合姿态。LoRA 负责质感，提示词负责信息密度，机位负责大胆度。 👇 点击上方卡片展开作者原图（含裸露内容 · 18+，再点收起） 用 §「快速开始」的命令可复刻淋浴那张（复刻输出已验证与作者原图同场景同构图，见 downloads/samples/phase5_meta.json 与本文快速开始一节）。 默认人物图 references/src_standing.png；姿势 = 骨骼旋转 JSON（MakeHuman 网格 + three.js…

项目简介：AI 自主渗透测试系统 | 百度“agent+”攻防挑战赛冠军项目 Autumn-27/ARTEX

技术栈：Go

核心介绍：AI 自主渗透测试系统（Go 后端 + Next.js 前端） 全局「审批记录」、任务内「拦截审批」及对话中的审批卡片均支持展开查看详情。展示结构参考 AegisHook 的审批详情组件，沿用 ARTEX 的组件和主题： 支持从 ScopeSentry 直接同步资产数据，免去重复收集：

项目数据：⭐ 1,198 Stars，🍴 11 Forks

项目简介：Open source canva alternative

技术栈：TypeScript

核心介绍：Doory is a design editor that runs in the browser: posters, social posts, stories, slides, invitations, resumes, thumbnails and business cards. You start from a template or a blank page, edit everything on it, and export PNG, JPG, SVG, PDF, MP4 or GIF. There is no paid plan, you don’t need an account to start, and you can run it on your own server with Docker or without it. It is a young project: the editor is complete, the template library is small (ten starter templates), and bug

项目数据：⭐ 1,196 Stars，🍴 69 Forks

项目简介：Getting ADB, root, backups and custom apps onto locked-down Android photo frames (Skylight Calendar, BIUFRAME)

技术栈：Python

核心介绍：Turn a cheap Android photo frame into a family dashboard: unlock it, root it, replace the vendor app, and run your own web page full screen. Frames confirmed to work (Amazon affiliate links). See the device index for every frame, which board it has, a

项目数据：⭐ 1,151 Stars，🍴 157 Forks

Tool-call judges evaluate argument values proposed by LLM agents. We find that they accept a correct value less often when their input holds no reference for it, that is, no text that says what the value should be. This happens when the value is a default value from the function schema that the user did not mention, when it has to be inferred, or when the text it comes from was cut from the judge input. We call this drop in acceptance the derivation gap. It appears across all seven judges we evaluate, including System One models CLM and Jev. The gap is not only a truncation artifact: correc…

Multi-robot systems (MRS) offer significant advantages over single-robot platforms by enabling information sharing and coordination for different tasks in complex environments. However, reliable cooperation critically depends on inferring trust over both information sources and their reported hypotheses, particularly under partial observability and potential adversarial interference. Existing approaches frequently depend on pre-established trust relationships or side-channel information, which may not be available or reliable in realistic MRS deployments. This work introduces NeST-Bayes, a …

Work agents operate over digital artifacts to execute professional knowledge-intensive work, requiring training environments that support long-horizon interaction and trustworthy verification. However, hand-crafted environments incur prohibitive engineering overhead that prevents environment scaling, whereas synthesis methods sacrifice workspace complexity, realism, or grounded verifiability. To bridge this gap, we introduce WorkForge, a scalable synthesis framework for constructing verifiable work-agent environments from real-world resources. Starting from expert workflows, WorkForge first…

Training on broad, unfiltered data enables high-quality image generation, but can also teach models concepts that should not be generated; restricting the training data ensures safety, but reduces quality because examples containing unsafe concepts also provide useful knowledge that is not itself unsafe. Existing approaches largely address this trade-off post hoc through concept unlearning, which can lead to catastrophic forgetting as the number of concepts to remove grows. We instead ask whether unsafe concept knowledge can be compartmentalized during pretraining so that access to it can b…

Intent-based network management automates 6G service lifecycles, but existing fulfilment and assurance approaches require traffic inspection, per-flow tracking, or vendor-internal telemetry. These requirements conflict with privacy constraints in multi-tenant networks and vendor opacity in O-RAN deployments. We present an architecture that provisions resources from declared intent categories without traffic inspection and verifies fulfilment using only cell-level standardised PM counters at the O1 interface. A data-processing inequality bounds resource-allocation leakage about traffic conte…

Latent diffusion models generate images as codes of an image tokenizer, whose decoder turns them into the pixel image. Recent tokenizers improve how well these codes reconstruct an image and how easily a generator learns them. Yet a generator does not reproduce the codes of training images, and when a generated code lands outside what the decoder expects, the decoded image contains imperfections. Current methods make the decoder more robust to generated codes by adding noise during training. However, generated codes lie between the training codes and resemble their mixture more than one of …