2026年6月27日 技术热点总结

项目简介：Codex 橙皮书：从安装到实战案例的全链路 Codex 使用指南（非官方开源，含可下载 PDF）

技术栈：HTML

项目数据：⭐ 2,161 Stars，🍴 219 Forks

项目简介：可溯源的郑希(易方达基金经理)投研 Agent Skill——基于他全部公开观点原文 + 有原话佐证的投资方法 + 全市场基金真实数据，能溯源问答、按他框架给基金打分，绝不杜撰。⚠️仅研究学习辅助，不构成投资建议‼️website是郑希主页！

技术栈：Python、agent-skill、chinese-funds、funds、investing

核心介绍：安装 · 它能做什么 · 在其他 AI 里用 · 目录结构 · 数据来源 · 边界 很多关于”某基金经理怎么看 X”的回答，听起来头头是道，其实是模型凭印象编的——查无此言。这个 skill 解决的就是这个问题：它的根基是郑希（易方达权益投资管理部副总经理、基金经理）本人留下的原始文本，任何结论都能追溯到”哪一年、哪一篇、原话是什么”。 它有三块根基，都来自公开内容、可溯源： 下面是 skill 的真实输出（稍作精简）。 ① 溯源问答 —— “郑希怎么看光通信？什么时候开始看好的？” > 他在 2026 年 6 月接受中国证券报采访时把逻辑讲得很清楚： > > “全球 AI 资本开支已经来到万亿美元级别。AI 产业链中，数据传输是重要一环，而光通信则是远距离传输的关键路径。叠加我国在光通信领域的全球比较优势，未来光通信市场规模有望进一步提升。”

**…

技术栈：JavaScript

核心介绍：No missions. No scores. No stress. A relaxing, browser-based bird sim where you steer a flock of seagulls in V-formation over the ocean — through day, night, storms, rain, and lightning. It’s not really a “game.” It’s a quiet, soothing experience. On mobile, use the on-screen buttons. Serve the folder with any static file server: python -m http.server 8000 Then open http://localhost:8000.

关键特性：Runs fully in the browser (WebGL / Three.js)；Physics-based flight with flock V-formation；Dynamic weather: day, night, storm, rain, lightning；Ambient music；Smooth on m…

项目简介：Arma: Cold War Assault Remastered Source Code Repository.

技术栈：C++

核心介绍：This repository holds the engine and game source code (codename *Poseidon*) behind *Arma: Cold War Assault* — the game first released in 2001 as *Operation Flashpoint: Cold War Crisis*. That release launched Bohemia Interactive and began the technology lineage that later grew into Real Virtuality, Arma, and Enfusion. The code has been modernized to C++20, built with CMake and Clang, with cross-platform support for Windows x64 and Linux x64.

项目数据：⭐ 634 Stars，🍴 71 Forks

项目简介：Qwen-AgentWorld: Language World Models for General Agents

技术栈：Python

核心介绍：📑 Technical Report | 📖 Blog | 🤗 Hugging Face | 🤖 ModelScope | 🖥️ Demo Welcome to the GitHub repository of Qwen-AgentWorld. Here, you can find official information about Qwen-AgentWorld, post your questions (Issues), and share your ideas with the community (Discussions). We open-source Qwen-AgentWorld-35B-A3B (model weights) and AgentWorldBench (evaluation benchmark): The official weights and data are released on:

项目数据：⭐ 573 Stars，🍴 51 Forks

项目简介：🏸 AI Badminton Hawk-Eye System

技术栈：Python

核心介绍：中文 | English 视频效果在 assets/demo.mp4。 参考速度会受显卡、视频分辨率、姿态模型、是否显示窗口、是否保留音频影响。 以 720p 视频、–pose-family yolo-pose –yolo-pose-model yolo11n-pose.pt 和 weights/yolo11s-ball.pt 为例，GPU 推理日志通常接近： pose 0.02s, shuttlecock 0.02s, shuttle draw 0.00s, players draw 0.01s, court draw 0.00s 开启 –performance-stats 可以每隔约 5 秒打印一次性能汇总，用于判断瓶颈在姿态推理、羽毛球检测还是绘制阶段。

关键特性：球员姿态检测 – 支持 RTMPose、RTMO 和 Ultralytics YOLO Pose，识别人体关键点和骨架。；羽毛球检测 – 使用 YOLO 模型检测羽毛球位置，并在输出视频中绘制轨迹。；球场坐标映射 – 手动标注球场关键点，将图像坐标映射到标准球场坐标。；球员位置追踪 – 分别追踪上半场和下半场球员位置，记录移动轨迹…

项目简介：修改 Apple 网络定位（gs-loc）返回坐标 · 支持 Surge / Quantumult X / Loon / Stash · 快捷指令一键设置/恢复定位

技术栈：JavaScript

核心介绍：修改 Apple 网络定位服务 (WiFi/基站) 返回的坐标，实现 iOS 网络定位虚拟定位。打开在线选点页面选位置即可生效，无需手动填经纬度。 > Egern 可直接使用 Surge 模块 > Stash 请直接订阅上面的 .stoverride，无需用 Script Hub 转换 直接用快捷指令切换 / 清除定位，无需打开选点页面： 支持苹果地图、高德（含短链，自动跟跳转 + GCJ-02→WGS84 坐标换算）。 > 前提：代理已开 + 模块已启用 + 信任 gs-loc.apple.com。选点页面（Worker / Pages）方案仍保留，见下方。 为了让苹果地图和高德走同一条流程，链接统一发给 wloc-spoofer.wloc.workers.dev/api/parse 解析： 使用方法 1. 订阅模块并启用 MITM 2. 打开在线选点页面（公共 Worker，建议添加到主屏幕） 3. 地图选位置 / 搜索地名 / 粘贴地图链接 4. 点击「储存到设备」 5. 下次 Apple 定位触发时自动生效

项目数据：…

项目简介：Eleven autonomous AI agents open live football prop markets on X Layer — custom Uniswap v4 hook, gasless USDT0 staking.

技术栈：TypeScript、ai-agents、defi、eip-3009、erc-8257、football、foundry、gasless、mcp

核心介绍：> Live football prop markets, made by AI agents. Built for the 2026 > tournament on X Layer. Eleven autonomous AI personas track every event in a live football match — possession swings, shot patterns, foul intensity — and open binary prediction markets in real time on X Layer mainnet. Each market is a custom Uniswap v4 hook that settles in USDT0. Users stake ga…

项目简介：A curated, non-BS library of the best resources for building and evaluating AI agents — papers, blogs, talks, tools, benchmarks. Maintained by BenchFlow.

技术栈：agent-evaluation、ai-agents、awesome、awesome-list、benchmarks、evals、llm、llm-evaluation

核心介绍：> A curated, opinionated, non-BS library of the best resources for building and evaluating AI agents — papers, blog posts, talks, courses, tools, and benchmarks. Maintained by BenchFlow · Most “awesome” lists are link dumps. This one is annotated and verified: every entry says *what it is and why it belongs*, URLs…

项目简介：“AgentSpace: Human + Agents. One Team. One Workspace”

技术栈：TypeScript

核心介绍：AgentSpace: Human + Agents. One Team. One Workspace English | 中文 AgentSpace brings humans and agents together — as one team, inside one workspace Feishu was built for humans. AgentSpace is built for both. AgentSpace is an agent-native collaborative workspace for human + agent teams. Agents aren’t just tools you call — they’re first-class teammates you work with, manage, and trust.

项目数据：⭐ 458 Stars，🍴 49 Forks

ABACUS is a unified vision-language model that handles object counting, crowd counting, referring-expression counting, and count-faithful image generation without any benchmark-specific training required. Our model is built on existing 3B-parameter unified foundation model and is adapted for object localization tasks using three key innovations: density-aware adaptive zooming with objectness maps for spatial grounding; a boundary-aware count policy via GRPO to eliminate crop-boundary errors; and a cycle-consistent GRPO strategy where the understanding branch self-critiques generated outputs…

Process reward models enable fine-grained, step-level evaluation of LLMs, yet building them for agentic settings remains prohibitively difficult: long-horizon interactions, irreversible actions, and stochastic environment feedback make both human annotation and Monte Carlo estimation infeasible at scale. In this work, we show that reinforcement learning (RL) post-training already provides the ingredients for effective step-level scoring, eliminating the need for dedicated reward model training altogether. Concretely, we derive an implicit advantage under a general stochastic Markov decision…

Reasoning capability has advanced rapidly in large language models (LLMs), leading to an increasing size of key-value (KV) cache in both prefilling and decoding stages. Existing KV cache compression methods mainly rely on attention weights to estimate token importance. While attention effectively captures contextual relevance, it overlooks complementary information-theoretic signals related to predictive uncertainty and token informativeness. In this paper, we revisit token importance from a forward-looking perspective and introduce Forward Influence, a metric that measures how compressed t…

Earth Observation (EO) forecasting aims to predict future Earth surface dynamics from satellite observations under changing meteorological conditions. In this paper, we view this task as a partially observed, weather-driven world modeling problem, in which weather acts as a conditioning signal, while forecasting remains uncertain due to sparse observations and unobserved land-surface states. However, existing methods do not fully capture this setting: deterministic models collapse uncertainty into a single future prediction, while diffusion-based methods typically treat weather variables as…

The prevalent dual-branch paradigm, i.e., training a side network to encode visual conditions and fusing its intermediate-layer features to a frozen pretrained main network, has shown remarkable success in visual-condition controllable generation. Despite its widespread adoption, the role of the side branch and its training efficiency remain underexplored. In this paper, we first revisit this mainstream paradigm through the lens of score-based generative modeling: 1) The main network preserves visual perceptual quality by providing a prior unconditional score. 2) The side network steers con…

As agentic systems continue to evolve and are widely deployed in real-world scenarios, there is a growing demand to faithfully evaluate their capabilities. However, current benchmarks are typically built on popular applications with relatively simple tasks and focus on a narrow set of capabilities while overlooking broader dimensions, resulting in saturated performance on modern agents and failing to probe their limitations. To this end, we introduce GauntletBench, a web-based benchmark for evaluating agent generalisation in challenging scenarios, focusing on three underexplored capabilitie…