The free launchpad. Uniswap v4 token launches with zero platform fee and permanently locked liquidity, on Base and Robinhood Chain.
CAP: Continuously Adaptive Perception-Blind Humanoid Locomotion via Learned Denoising
Telegram file uploader bot powered by Cloudflare Workers
Nothing OS look for Hyprland: Quickshell bar/control-center/dock/widgets, Nothing motion, monochrome icons, wallpapers, SDDM + Plymouth themes. Built on JaKooLit's dots.
Browser extension that auto-fixes Persian typos as you type
Un conto ilustrado para aprender sobre licenzas, comunidades e contribucións Open Source.
Jev-powered, rule-based grader for text files. Runs every rule against every line in parallel. No skimming, no missed lines.
—
Let Jev read every new arXiv paper each morning and surface the few you should read. Plain-English interests, calibrated probabilities, ~$0.06/day, fork and go.
Edite vídeos conversando com o Claude: Nível 1 gratuito (Remotion) e Nível 2 avançado (Higgsfield).
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
A curated collection of advanced Prompt Engineering templates for LLMs to optimize AI workflows and productivity.
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
DSH host RSI for local LLMs (currently Qwen3.8 27B): per-use continual self-improvement, counter-reason self-check, critic escalation, and memory retrieval.简单来说根据论文建立的长期反思教训机制,让大模型不再犯错,越用越强!重点是千问3.8 27B/FLASH这种本地LLM模型!当然在线大模型理论上也可以通用!
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
Experimental Windows and Linux only passthrough mod that lets you play Team Fortress 2 inside Minecraft Java using Minecraft’s world, blocks, mobs, and terrain. Made using Claude x ChatGPT.
Nintendo 64 emulator (mupen64plus-core 2.6.0) as a native PS5 home-screen app: dynarec, DualSense rumble, cover shelf, save states, auto-updates
Custom film looks (Ektar, Gold, Portra, Pro 400H, Vision3 250D) and RAW-NR mods for the Ricoh GR III firmware 2.10 - tools and research notes
麦麦餐厅-慧眼识人 | 麦当劳 2026 年 1024 程序员创意开发大赛参赛项目:基于麦当劳 MCP 真实餐品数据的出餐配餐与优惠推荐练习游戏(单文件 HTML,双击即玩)
Fan remake of the classic RTS Tzar: The Burden of the Crown, rebuilt in Rust and Bevy. Needs your own copy of the game (GOG, Steam or CD).
—
—
—
Open Source implementation of xgameruntime.dll. Designed for use in Wine with Xodus
velocity but improved for pasters
A collaborative tool for RFCs. Complete slop
DSH Web UI 右侧小游戏面板:18 款离线小游戏(恐龙跳一跳 / 俄罗斯方块 / 坦克大战 / 扫雷 / 2048 / 数独 / 吃豆人 / 跟枪练习等),可扩展游戏注册表,等待模型回复或修 bug 时的摸鱼神器
An unofficial macOS launcher for the Global PC version of Arknights
Unofficial Android Studio install for AArch64 Linux (ARM64)
Deep reading & summarization workflow for DSH: books/papers/videos/web → structured Markdown notes. Plugin parsers, MapReduce deep-read, JSON Schema output.
A massively performant superset of the Figma MCP surface, ensuring your swarms of coding agents are never rate limited. 17,657× faster than Figma. Written in Rust and built on Flower Computer's Bog db toolkit https://github.com/flowercomputers/bogkit
—
一个Turing专用vLLM分支,通过整合FlashInfer 0.6.18(PR #3526)与SM75后端门槛补丁(PR #47949),持续同步vLLM主线最新版本
Any track becomes a game soundtrack: AI-detected sections, seamless loops, beat-aligned transitions, intensity layers.
CloudStream Beta için 56 ayrı film, dizi, anime ve belgesel sağlayıcısı; !turksinema.
Dashlaunch addon to allow wired Xbox One/Series controllers directly with no dongles.
A suite of AI agent skills for planning, designing, validating, and building credible B2B websites with maintainable custom Elementor widgets.
A directory of benchmarks for the components and complete systems used to build voice agents.
DLSS 5 neural rendering overlay and video processing for Apple Silicon Macs.
Terminal fleet monitor for GPUs, NPUs, and 13 other accelerator kinds, from utilization down to which pod is using them.
看看 Jev 能做什么:用中英文讲清热门应用、工作原理和各自优缺点。Explore Jev apps with plain-language examples, explanations, and comparisons.
—
Community Components for Rocket: a gallery of Datastar Rocket web components. Go, templ, SQLite, CQRS.
—
Study trackers, Obsidian prep vaults, practice boxes and resources for Hack The Box certs: CJCA, CPTS, CWES, CDSA, CWEE, CAPE, CWPE, COAE and CDEP. Each cert gets a day-by-day plan, a ready-made notes vault, a list of HTB Labs boxes to practice on, and every study resource in one place. Trackers work offline. Not affiliated with HTB.
视频精简:腾讯视频、优酷、爱奇艺广告拦截与界面精简 LSPosed 模块
Make a complete drama song ad: a long-form ad sung as a drama, from script to song to finished video. Originally designed for the MaxFusion AI MCP, tool-agnostic: works with any image, audio and video generation tools, or by hand.
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
AI API gateway / relay: 307 models behind one OpenAI-compatible key. image2.5 $0.0085/image, Seedance 2.0 Mini $0.01056/sec, LLM from $0.0228/M tokens. USD settlement, pay-as-you-go, $1 minimum top-up.
Triage de seguridad para Windows (procesos, red, tareas programadas, persistencia, Windows Defender) con veredicto en lenguaje simple. 100% portable, open-source friendly, creada por @kobasky para su comunidad.
WHIRL (Windows HIP Inference for RDNA LLMs): native Windows C++/HIP LLM inference engine for the AMD Ryzen AI Max+ 395 (Radeon 8060S) and Radeon AI PRO R9700. Unzip and run, only the AMD driver.
A calm, fast Kubernetes dashboard, on your desktop or in your cluster.
—
LightVAE: Fast video VAE decoders for Wan2.1, Wan2.2, and MiniMax-H3, plus a compressed H3 reference encoder. Pretrained weights and ComfyUI workflows.
Windows本地bonsai模型部署启动器
—
Experimental NFSU2 VR build for Windows PCVR and Meta Quest 3 standalone. DEV BUILD: not fully tested; bugs and performance issues may occur. No game content included.
Interactive terminal client to reclaim gigabytes on your Immich server by re-encoding videos to hardware-accelerated HEVC while preserving 100% of metadata