2026年7月21日 技术热点总结

项目简介：Rebuild the object in a reference image as a code-only, procedural, quality-gated, animation-ready Three.js model. Token-efficient image-to-3D.

技术栈：Python、3d、ai-agents、claude-code、computer-graphics、generative、image-to-3d、procedural-generation、threejs

核心介绍：Quality-gated, animation-ready, and deliberately token-efficient — reconstruction-by-code, not photogrammetry, mesh extraction, or downloaded art packs. Every model in the gallery is generated code, running in your browser. No mesh files, no downloads. Reconstructions built entirely from primitives, procedural shaders, a…

项目简介：Use any agent to create stunning, interactive presentations

技术栈：TypeScript

核心介绍：One prompt in — your agent builds a deck where every slide is a live, responsive web page. 3D, live data, working prototypes, and whatever you can prompt. AI for slides is awesome, but the outputs tend to be slop: generic layouts, walls of bullets, nothing you’d be proud to present. And also: why are slides still *static*? Agents can build *anything*. What would it look like if you (tastefully) turned them loose on slides?

项目数据：⭐ 501 Stars，🍴 62 Forks

项目简介：Turn hand-drawn story illustrations into 35–45 second line-reveal and gradual-coloring videos with HyperFrames.

技术栈：Python

核心介绍：把 7–9 幅手绘故事母图制作成 35–45 秒竖屏短视频：先从左到右显现黑白线稿，再沿相同方向逐步填入低饱和色彩。项目包含故事配置校验、线稿提取、HyperFrames 页面生成，以及可安装的 Codex Skill。 新手直接看：中文使用方法 · 生图模型接入 · 固定画风提示词 想直接按主题批量生成：一键成片配置 git clone https://github.com/xiejunjie524/handdraw-story-video.git cd handdraw-story-video pip install -r requirements.txt 1. 复制模板并填写故事： cp templates/story-template.json story.json 2. 每幕只生成一张 1K 彩色母图，再提取对齐线稿： python scripts/make_lineart.py assets/images/scene-01-color.png assets/images/sc…

项目简介：Circuit Framework — multi-agent LLM trading research system

技术栈：Python

核心介绍：A crypto-native multi-agent research and paper-trading framework where specialized agents analyze market structure, derivatives, sentiment, catalysts and market regime before a deterministic risk engine approves or rejects each trade. Circuit Framework is a fork of TradingAgents (Apache 2.0). The internal Python package remains tradingagents for compatibility; user-facing branding, CLI help and crypto workflows are Circuit Framework.

项目数据：⭐ 486 Stars，🍴 19 Forks

项目简介：半调纸拼贴 B-roll 生成 skill：三闸门审批，Gemini Omni Flash 首尾帧组装动画 | Editorial halftone paper-collage B-roll agent skill

技术栈：Python、agent-skill、broll、gemini、paper-collage、video-generation

核心介绍：把一句约 5 秒的口播文稿，压成一个 sharp visual idea，再生成高级编辑风半调纸拼贴（halftone paper-collage）组装动画 B-roll。 Turn a ~5s voiceover line into a premium editorial paper-collage assemble-from-empty B-roll clip, powered by Gemini Omni Flash first/last-frame video generation. 这个 skill 的核心不是 prompt 模板，而是强制的三阶段审批，让你把注意力花在审美判断上，而不是烧生成费用： 1. Gate 1 · 隐喻确认 — 只输出视觉隐喻方案（核心意思 / 关键物件 / 底色 / 组装顺序），不生成任何图片视频 2. **Gate 2 · 静帧确…

Security-agent evaluations commonly measure peak offensive capability under generous inference budgets, emphasizing vulnerability discovery, exploit development, penetration testing, and CTF completion. Such measurements are useful but incomplete: in operational security, every reasoning step, tool call, telemetry query, and enrichment request consumes budget. We evaluate language-model security agents through this cost-success lens on offensive Cybench challenges and defensive Splunk BOTS v1 investigation challenges. Instead of reporting only best-case success, we compare models at fixed c…

We introduce Self-Verified Reasoner (SVR-R1), a multi-turn RL framework that turns a model's own verification into a learning signal for multimodal reasoning. For each query, the model proposes an answer using the same weights, and issues a binary self-verdict (Yes/No). A 'No' triggers a second-chance rethink; a 'Yes,' or a turn cap, finalizes the output for computing the outcome-based reward. SVR-R1 is implemented with GRPO and an asynchronous multi-turn rollout framework and needs no external supervision or auxiliary critics. We evaluate SVR-R1 on vision-language reasoning benchmarks and …

Plasma diagnostic models for tokamak fusion devices are almost universally evaluated on clean, complete sensor data. In practice, fusion diagnostics fail regularly: acquisition systems start late, individual sensors die, and signal dropouts cluster precisely when a plasma disruption is approaching. We present the first systematic robustness benchmark for plasma diagnostic ML using the TokaMark dataset of 11,573 MAST shots, evaluating XGBoost, LSTM, Transformer, and the TokaMark CNN baseline across six physically-grounded failure scenarios and three imputation strategies. We introduce the Ro…

Training-free in-context segmentation enables new object categories to be introduced at inference time from a single annotated reference image, eliminating the retraining and memory overhead of class-incremental learning. Recent approaches achieve this by combining vision foundation models for semantic correspondence with promptable segmentation networks like SAM. However, their performance is fundamentally limited by the quality of the cross-image similarity map; shared contextual backgrounds between the reference and query systematically elevate similarity in non-target regions, degrading…

Autonomous negotiation agents are increasingly deployed in high-stakes settings such as insurance and procurement. While cryptographic techniques protect explicitly disclosed constraint values, they fail to address a subtler threat: behavioral privacy leakage, where an adversary infers private constraints from observable negotiation dynamics such as concession trajectories, timing, and convergence patterns. This paper investigates behavioral differential privacy in multi-round negotiation protocols. We design an adaptive stochastic negotiation policy that jointly guarantees (varepsilon, δ)-…

Vision-language-action (VLA) models predict robot actions from visual observations and language instructions. These actions are defined in the robot's own 3D coordinate frame, yet most VLAs observe the scene in the camera frame, creating a frame mismatch between where the scene is observed and where actions are defined. The mismatch is benign under a fixed viewpoint, where the policy can memorize a single observation-to-action mapping, but grows harder as large-scale datasets aggregate demonstrations across diverse camera setups and the policy must generalize this mapping across viewpoints….