Make use of your old phone || Your own Desk Buddy
HRFFA: High-Angle Robust Fast Face Alignment. Whole-head face alignment dataset & training pipeline.
Persian text-to-speech with voice cloning, fully offline on CPU (ONNX + torch paths)
Open recipes, engine patches, and benchmark harnesses for LLM inference on Intel Arc Pro B60/B70 (Battlemage, Xe2). MoE 35B at 160 t/s decode / 7.5K t/s prefill single-stream, 27B at 50~ t/s decode / 1.7K t/s prefill single stream. vLLM XPU MTP unlocked. Muse Glimmer recipe added!!
Biome-S1: a A family of tiny System-1 executors for computer-use agents
软件著作权 Skill,自动生成软著申请表、用户手册、技术文档及代码材料,辅助完成软著登记准备。
—
Official code implementation of "TERRA: Terrain-Aware Reconstruction, Retargeting and Control for Musculoskeletal Locomotion"
A model for high-frequency mathematical intelligence.
MiniMax H3 驱动的全自动 AI 视频导演流水线——自动分镜、工作流分配、参数确认、批量生成、剪映一键拼合。
过审 guoshen:联合语音、字幕与画面,按小红书、抖音、B站、视频号动态规则预检;开放案例与社区共建。
Local-first recruitment intelligence and application operations agent
Multi-agent AI system for drug-target identification and due diligence.
Use local Qwen3.8-27B and official MiniMax-H3 Skills in ComfyUI to generate H3 prompts. Runs a standalone local llama-server , inference operates fully offline with no internet access.
Malware as a Service Ludus Range
Air traffic control discipline for agent output. Written for a reader with ADHD.
Autonomous multi-agent pentest framework — plans, exploits, verifies (proof-required), CVSS-scores and writes client-ready reports. 104/104 (100%) on the XBOW validation benchmarks, powered by Kimi K3.
把 SRT 字幕变成 Vox 风格解释视频的分镜与提示词包:21 种风格库、五模式流水线、按 4/6/8/10 秒吸附的分镜算法,只交付提示词,不生成媒体。
Long-horizon benchmark where 18 LLM agents got ¥100,000 each and ran simulated online stores for 365 days on real market data: negotiating with suppliers, pricing, managing inventory, keeping cash flow alive.
Blacksea is an active honeypot and canary-bait control system built to detect and drown LLM-driven attackers. Blacksea doesn't stop at watching LLM attacks. It exploits flaws in the attacker's LLM judgment to gain arbitrary code execution on their machines, collect intel passive defenses can't reach, and make sure they don't come back.
—
AIBuildAI LLM-Post-Train Agent: a recursive self-improving (RSI) agent for autonomous LLM post-training
知乎 AI 蓝宝书 · 三册文稿(AI Infra / RL / 具身智能)
Pretrained text classifiers you can run and retrain on CPU
[NeurIPS 26] Official implementation of the paper "Urbex: Agentic Spatial Grounding for City-Scale 3D Scenes"
一个将人物图片、参考声音和口播稿转化为完整数字人口播视频的开源 Agent Skill。
Rethinking World-Action Model for Compositional and In-context Robotic Manipulation
首个MiniMaxH3 ComfyUI加速插件
AI agent operating system for Blender 5.2: bpy knowledge base, AGENT_OK execution contract, 3D-print production gate, worked keyboard and desktop-robot builds.
延续一个拟物设计时代留下的精致痕迹
—
一套写给 agent 读的求职方法论。不是投递脚本,是每个决策点上的判断纪律。
AI fight-video creation/design skill | 从打斗构思到剧情、动作、分镜;支持资料检索、招式校验与参考库吸收。
小红书养号工具:拟人采集·评论·点赞·搜索
Проверь свой навайбкоженный проект на дыры до того, как это сделают боты 🩺
Un journal personnel A4 genere chaque matin pour remplacer le scroll.
Django tasks in your database. No Redis or RabbitMQ. Unfinished work returns to the queue when workers die.
evoX-CoreOS
Homework assignments for Evaluating and Improving AI Agents
Carveout finds and labels objects in trained 3D Gaussian Splatting scenes.
—
蓝色星原:旅谣自动化脚本
MiniMax H3 300% speed本地部署指南(消费级 16GB 显存版):全栈加速链 cu130+SageAttention+TE-Speed 实测、模型组合、避坑清单
A local acoustic control system that learns user-defined labels for the surface or location of knocks, such as desk or wall. It also counts threshold-qualified knocks within each captured window. By default, a frozen Microsoft BEATs Iter3+ AS2M encoder creates 768-D audio embeddings and an incremental classifier learns the user's labels.
Official code for "Relevance Does Not Imply Applicability: Experience Activation for Personal GUI Agents".
白板讲解动画 skill。把讲解稿变成「真手绘过程」视频——图形一笔笔描出、文字逐笔写出、讲到哪画到哪(帧级音画对齐)。 > 本 skill 是**像素级笔画绘制**——帧帧确定、本地零成本、可秒级重渲,且支持"一图逐元素生长"的单画布模式(数学推导/流程图 友好)。
Fast, flexible, elastically deterministic data loading for foundation model training
Googlebook OS on x86_64 PC Hardware
本地 ComfyUI 抽卡引擎,解决不会写提示词的问题。
—
AI 校园版《败犬女主太多了!》OP 制作过程:分镜、MiniMax H3 逐镜人物替换、拼接、超分;18 位 AI 娘人设与完整演员表
Find people publicly describing the problem you solve. Local lead-signal listener for Reddit, X, LinkedIn, HN, RSS and 3,800+ Treg endpoints — scored, de-duplicated, bilingual UI. By @yangyi · built by newmax.ai
A fast, multi-format document parsing and conversion engine
Recursive Game Creator: An Agentic Product-Level Experience-Oriented Game Harness
DeepCloud 桌面客户端发行版发布仓库 (Releases & Distribution)
—
Windows AI companion with DSH long-term memory, local voice cloning and optional WSL2 digital human.
Reconstructing DX-Ball 1.07 with REA, backed by original-x86 differential tests and pinned MSVC byte-exact oracles.
Local decision engine with typed, calibrated outputs (Noul/Choice/Score): cheap→strong cascade, ECE/Brier calibration, SystemOne protocol interop. Apache-2.0.
A composable framework for reproducible evaluation of long-running agents.