09.12 Today's Picks

320 new projects found today, 174 rated.

Created 2026-09-10 – 2026-09-12Fetched 600Filtered 50Rated 34

A · Worth watching

tracecrate

FankChenTypeScript

A local-first AI agent trace workbench that reads Claude Code, Codex and OTLP logs, compares runs and exports redacted reports, targeting agent debugging and compliance pain. It hit 98 stars in 1.6 days with no backend or keys needed, and could monetize via team subscriptions and enterprise audits; its main rivals are cloud observability platforms like Langfuse and LangSmith.

Local-first AI agent trace workbench. Inspect Claude Code, Codex and OTLP logs, compare runs, and export privacy-conscious reports. No backend or API keys.

#AI Agents #Observability #Debugging #Privacy

★98

reburp

forefyKotlin

Exposes Burp's entire Montoya API as a local REST interface with Swagger UI, letting scripts and agents drive Burp for automated testing, filling a real gap. Clear scope and an obvious starting path; concurrency and stability still need real-world testing.

A Burp Suite extension that exposes the full Montoya API as a local REST API, with Swagger UI

#App Security #APIs #Plugins

★46

crossmatch

filip131311TypeScript

Drives both the iOS and Android versions of an app and records each UI difference as side-by-side video, filling the neglected gap of cross-platform consistency testing. Clever idea; the hard part is running both automation suites on real devices, and setup instructions are still lacking.

Find where your iOS and Android apps disagree, and get a side-by-side video of every difference.

#Testing #iOS #Android #Video

★22

Toneferry

magicflower2JavaScript

Streams Windows system audio to a phone on the LAN for real-time listening—one sentence explains the use and it works once installed, good for people who want to hear their PC on their phone. Closed-source tools like AudioRelay exist, but it gives no latency or audio-quality data.

Windows system audio → real-time listening on a phone over LAN.

#Audio #Streaming #Windows

★12

askWAM

dirkjanmC

A small tool that requests Entra ID tokens directly from Windows Web Account Manager, from a security researcher familiar with the mechanism, saving red teams and token forensics a lot of guesswork. Clear scope and restrained description; only meaningful in authorized testing scenarios.

Ask the Web Account Manager (WAM) for Entra ID tokens

#App Security #Windows #Secrets

★12

sctxx

handyutilsRust

Extracts session records from different agents like Claude Code and Codex into compact handoff files with source links, so another agent can continue the work; cross-agent context migration is indeed a new problem. The concept is clear; compression quality and format changes across vendors are the main risks.

sctxx — coding agent Session Context Extractor. standalone, agent-agnostic CLI (plus a thin Agent Skill wrapper) that reads a coding-agent session transcript from disk — Claude Code, Codex CLI, or Pi …

#AI Agents #Agent Skills #CLI

★18

foldelight

lufzleSwift

Drives a native macOS origami-glass animation from a MacBook's lid angle—tiny in scope and restrained in description, the kind of toy you want to tweak a couple of lines of. Beyond fun it has no real use, and it's locked to a specific model's sensor.

A native macOS folding-glass effect controlled by the MacBook lid.

#macOS #Animation #Hardware

★16

catalogflow

qwq77777764Python

A product catalog pipeline for WooCommerce that emphasizes human confirmation and safety, putting a gate on the risky task of AI bulk-editing product data. The scope is clear, but it's only useful within the WooCommerce ecosystem.

Safety-first, human-approved product catalog pipeline for WooCommerce

#E-commerce #Automation #AI

★26

agent-launcher

agent-launchTypeScript

Wraps existing coding agent CLIs like Claude Code and Codex into one desktop shell for unified configuration and launching, solving a real hassle for people running several agents on one machine. Restrained in scope without hype; the challenge is keeping up with frequently changing CLI arguments.

Configure and run existing coding-agent CLIs from one desktop app.

#AI Agents #CLI #Automation

★27

Daub

directwirePython

Renders brush stroke JSON directly to PNG and outputs layered kra/psd files plus drawing replay videos, running batch renders outside the Krita process, useful for drawing-process videos or asset pipelines. The repo is tagged Python while the text says pure Rust, an inconsistency that needs clarifying first.

Headless calibrated-brush renderer: strokes JSON to PNG + layered .kra + layered .psd + reveal video. Pure Rust, no Krita process needed.

#Images #Video #Animation

★25

RomMMobile

pippopaperaKotlin

An Android client for people running their own RomM, downloading ROMs straight into the frontend's directory and saving a genuinely annoying step. The scope is clear and the description honest, but it's entirely tied to the RomM ecosystem, so the audience is narrow.

Android companion app for RomM: browse your self-hosted library and download ROMs straight into the right folder of the frontend you already use.

#Android #Self-Hosted #Games

★10

DuoLauncher

jakesgoodappsKotlin

An experimental Android launcher for foldables: a right-side dock, native widgets, and stackable home screens when unfolded, addressing the long-standing gap in Android large-screen adaptation. Fairly complete, but stability, battery drain, and support for new models remain to be seen.

An experimental Android launcher for foldables, with a right-side dock, native widgets, and overlapping unfolded Home pages.

#Android #UI Design #Productivity

★24

windows-wesp-research

xusheng6PowerShell

Reverse engineering and hands-on validation of a Windows endpoint security platform preview, closely following Microsoft's new interfaces and directly useful to EDR and offensive/defensive researchers; the limitation is that it's research notes, not packaged as a reusable tool.

Reverse engineering and live validation of the Windows Endpoint Security Platform preview

#Reverse Engineering #App Security #Windows

★10

That's today. The other tiers and the boards follow.

BHas a spark37

B flash.tmux Adds easymotion-style jump labels to tmux panes, written in Zig, single-purpose, and ready after install. Similar tmux-jump exists, but a lighter implementation is still worth trying; 13 stars mean it hasn't spread, and usage examples are missing. #Terminal #CLI #Productivity

Navigate terminal panes by jumping

and-rsZig★13
B MotionClone Reconstructs a reference video into an editable motion graphics project file; the video-to-timeline path is rarely attempted and appeals to people making title templates. Currently only a local Windows app, so reconstruction fidelity and output project format are what most need validation first. #Video Editing #Video #Animation #Windows

Rebuild reference videos as editable motion projects. Local Windows app.

blixvipPython★20
B emibios A GBA BIOS rewrite in Zig to replace the original firmware, a genuinely foundational piece for emulator development and homebrew cartridges. From a well-known reverse engineer with a clear scope; the limitation is a tiny audience, and compatibility can only be proven on real hardware. #Reverse Engineering #Embedded #gba

a GBA BIOS replacement

coolbho3kZig★12
B esphome-airplay2 An ESPHome component that turns an ESP32 into an AirPlay 2 receiver, building on an existing port and saving DIY audio box makers from debugging from scratch. Note the non-commercial license; latency and stability still need real-world testing. #IoT #Audio #Embedded

ESPHome external component that turns an ESP32 into an AirPlay 2 receiver. Non-Commercial licensed (port of rbouteiller/airplay-esp32).

henrikliedC++★16
B ClaudeStatsBar Shows current context usage and the approximate cost of the next turn in the Claude Code status bar, so you see cost before typing. Very practical for heavy users and restrained in scope; it relies on official usage figures, so model changes may skew it. #LLMs #CLI #Terminal

A Claude Code status line showing how big your session has got and what the next turn costs before you type

Field-Logic-LtdPython★17
B local-workspace-mcp A self-hosted MCP service that exposes local files, terminal processes, and Office documents to models in one place, saving you from installing a separate server per resource type. The space is already crowded (filesystem, desktop-commander), so differentiation must come from permission boundaries, which remain unclear. #MCP #Self-Hosted #AI Agents #Documents

Self-hosted MCP for local files, terminal processes, Office documents and named computers

arumwuPython★25
B eqbyear Uses swept sine tones so you can manually mark the resonant peaks of headphones or earbuds, then directly generates a parametric EQ. Very practical for audiophiles without measurement gear, works right in the browser; novelty is average, but it explains one thing clearly. #Audio #Web Frontend #Productivity

EQ headphones and IEMs by ear. Sine sweep, mark the peaks, get a parametric EQ.

DMS3tvHTML★28
B Devspeak A Claude Code plugin for Brazilian developers to practice technical English, role-playing a fictional international team with their own real code as material—a specific angle. But it's essentially prompt orchestration, the results depend entirely on the model, and it's hard to verify. #Agent Skills #Prompting #Learning

A Claude Code plugin that coaches Brazilian developers on technical English for international remote jobs through role-play with a fictional international team, practicing on your own real code.

rodrigoscharpJavaScript★14
B cubism-api-bridge An unofficial TS/Python bridge and local HTTP service for Cubism Editor's external API—tools like this are genuinely missing in the Live2D ecosystem, and the author honestly labels it experimental. The audience is very narrow, and interface stability can only be found by trying. #SDKs #APIs #Animation

Unofficial TypeScript/Python bridge and local HTTP server for Cubism Editor External API, with StandRig integration. Experimental.

sayaka-aiartTypeScript★10
B MiniWatts It uses private iOS APIs to read real charging wattage, input current and thirty temperature sensors, making a professional gauge for iPhone battery and thermal management. 51 stars in 1.4 days show geeks have real demand for raw battery data. Commercialization is constrained because private APIs cannot pass App Store review, leaving only sideloading; the risk is OS version changes killing it, with official battery health and CoconutBattery as competitors. #iOS #Hardware #Performance

An iPhone battery, charging and thermal instrument built on private iOS APIs — real charger watts, what actually reaches the cell, all 30 temperature sensors, and the USB-PD handshake.

ResistanceToSwift★51
B reelbench-skills A collection of study notes and tool skills around AI video organization, callable directly by agents like Claude Code to lower trial-and-error costs in video generation workflows. 162 stars in 0.8 days hits both the AI video and agent skill trends; it could evolve into a video generation benchmark or skill marketplace. The risk is that note-style repos are easily forked and split, lacking a maintainable engineering core. #AI #Video #Agent Skills

Learning notes and tooling skills for AI video - AI 视频相关的学习与工具 skill.

eternityspringJavaScript★162
B html-diagram A Claude Code skill that produces directly editable HTML charts, with a nice 'the document is the editor' angle; but the docs are mostly Korean, and similar chart-generation approaches are already plentiful, so the differentiation isn't strong enough. #Agent Skills #Documents #Web Frontend

A Claude Code skill for creating editable HTML diagrams. The document itself is the editor.

tonywjsHTML★38
B sloppy Static analysis for Laravel that specifically catches bad smells left by AI coding agents, running deterministically and locally without calling models. A fresher angle than another linter, honestly described; currently Laravel-only, so the audience is limited. #Code Quality #AI Agents #Automation

Laravel-aware static analysis for the code-quality patterns AI coding agents leave behind. Deterministic, local, no LLM.

HeyosseusPHP★11
B ai-data-extractor Exports conversation history from Claude Code, Cursor, Windsurf, and similar tools into structured data—a need anyone switching tools will hit. But each vendor's format changes often, making such extractors costly to maintain, and similar scripts already exist. #LLMs #Data Analysis #CLI

Free open-source extractor for AI coding assistant chat histories. Supports Claude Code, Cursor, Windsurf, Aider, Cline/Roo Code, and more.

kruzovic7Python★17
B knot A self-hosted bot that stores LINE text, images, and files into your own Cloudflare D1/R2, genuinely useful for heavy LINE users with clear boundaries; it's just deeply tied to the full Cloudflare stack. #Self-Hosted #Social #Databases

A self-hosted LINE memory bot for saving text, photos, videos, audio, and files to your own Cloudflare D1 and R2.

vendyluoTypeScript★12
B playstunts Rebuilds the 1991 Stunts / 4D Sports Driving in the browser with TypeScript, requiring players to supply the original game files, a fairly restrained approach to copyright. Old racing fans can play right away; the novelty is mainly the porting work itself, since gameplay and assets come from the original. #Games #Game Porting #Web Frontend #stunts

Native browser reconstruction of Stunts / 4D Sports Driving, with tools to build from your own original game files.

ACatWithEbolaTypeScript★12
B repkgs Mic92's experimental package set on native Nix: cheap evaluation, relocatable outputs, a nushell builder, and LLVM-only cross-compilation. Technically dense from a veteran author, worth reading for anyone pushing Nix's build model; documentation is thin, so you'll read the code. #Build #Linux #DevOps #nix

A package set on stock Nix: cheap eval, relocatable outputs, nushell builders, LLVM-only cross, compile cache in the compiler driver

Mic92Nushell★34
B worktree-import-guard Checks whether tests actually run against the source you think, since git worktrees plus editable installs easily read stale paths. A small but real idea, usable by adding one check; the audience is limited to people with multiple worktrees open. #Testing #Debugging #Code Quality

Verify that your tests actually executed the source tree you think they did.

StatXzy7Python★11
B duo-open Recreates the iPhone's frosted-glass crease effect on Android foldables, driven by real hinge angle and requiring no root—a clever idea. But it's just a visual easter egg, locked to specific models and OS versions, with limited practical value. #Android #Animation #Hardware

iPhone-Duo frosted-glass fold effect for Android foldables, driven by the real hinge angle. No root.

marcoazeemKotlin★15
B openeval A typed prompt-and-judge SDK with a CLI and results viewer for agent evaluation. The engineering looks clean, but mature tools like promptfoo already own this space, so switching only makes sense if you really care about types and evidence retention. #LLMs #Testing #Benchmarks #AI Agents

Typed prompt-and-judge SDK, CLI, and evidence viewer for agent evaluations

HonaTypeScript★35
B NotchOTP A native macOS app that puts TOTP codes under the notch, summoned with a hotkey, signed and notarized so installation is painless. Menu-bar TOTP tools are plentiful; its only selling point is the notch position, but it's small and complete. #macOS #Secrets #Productivity

Native macOS TOTP authenticator: your two-factor codes in a panel below the notch, one shortcut away. Developer ID signed and notarized.

Photon94Swift★9
B web-cli Reuses a logged-in Chrome session for local browser automation, callable by AI assistants and scripts—a practical idea. But Playwright, browser-use, and similar abound, and it doesn't explain how it isolates login-state risk or handles concurrency. #Automation #CLI #MCP

Local browser automation for AI assistants and scripts. Reuse your logged-in Chrome session and run ordinary tasks in the background.

selenaking2025JavaScript★18
B xjvpn A Mihomo-based Android transparent proxy client with per-app routing and domain audit logs, somewhat useful for root users troubleshooting phone traffic. But mature Mihomo clients like ClashMetaForAndroid and FlClash already exist; this just adds an audit layer, requires root, and targets a narrow audience. #Networking #Android #Privacy

Root Android transparent proxy client powered by Mihomo, with per-app routing and domain auditing.

yizhiyonggangdexiaojiaJava★19
B SwiftDependencyGraphExplorer Pick a Swift type and it locally draws its type-level dependency graph, exporting Mermaid or SVG, with source never leaving the machine. Useful for people digging through large Swift codebases; general dependency tools exist, but type-level granularity is the differentiator. #macOS #Code Quality #Debugging

A native macOS app for exploring type-level dependencies in Swift projects. Pick a folder, select a root type, and visualize its dependency graph with Mermaid — fully local, no source code ever leaves…

GlitchLoomSwift★10
B DuoFlip A menu bar utility that switches desktops based on the Mac screen's lid angle, doing one thing and explaining it clearly. It is fun for people who like gesture-based control; the limits are obvious too, it only serves macOS, and the actual feel depends on the hinge sensor. #macOS #Productivity #Hardware

Native macOS menu bar app with lid-angle-driven desktop transitions.

AresNingSwift★10
B decky A DIY 7-inch ESP32-S3 touchscreen Stream Deck, with both firmware and Windows software open source. Hardware tinkerers can build it; several open-source decks already exist, but this one wins on screen and feel details. #Hardware #Embedded #Windows #Productivity

Open-source DIY stream deck: touch firmware for the CrowPanel 7-inch ESP32-S3 and a Windows desktop app

FenishTypeScript★14
B image-gen-open-source A local workbench for generating, editing, and comparing images, with your own key connecting to OpenAI and Google models. The feature mix is practical, but such wrapper frontends are already plentiful, and "open source" doesn't include the models themselves, so the differentiation is unclear. #Image Generation #AI #Images

A local image workspace for generating, editing, and comparing AI images with Codex, OpenAI, and Google. Bring your own accounts and API keys.

koppkvnJavaScript★17
B better-context-engine Yet another codebase indexing and context retrieval engine for AI coding assistants, heavily overlapping with ACE, Augment, and similar. The description gives no retrieval-quality comparison or integration steps, so it's hard to tell what it does better than existing tools. #RAG #LLMs #APIs

ACE-compatible code context retrieval engine: real-time codebase indexing and low-latency context for AI coding assistants

linqiu919Go★21
B nadir-core An open-source library that scores ADAS sensor drift, integrated in shadow mode under Apache license—it targets a real engineering problem. But it gives no metric definitions or sample data, so outsiders can hardly judge its accuracy. #Robotics #Hardware #Monitoring

Open-source ADAS sensor drift scoring (Pulse/Echo). Shadow mode. Apache-2.0.

thedhruvhegdePython★15
B module-ballerina-a2a Ballerina's official implementation of the A2A protocol; the spec mapping is tidy, but the audience is limited to the few people writing agent services in Ballerina. A2A SDKs already exist for many languages, so this fills out an ecosystem rather than breaking new ground. #AI Agents #SDKs #APIs

Ballerina Implementation of the A2A Specification

ballerina-platformBallerina★10
B sk-music A Rust music player built on the Soulseek network—wiring a P2P protocol into a player is interesting, but clients like Nicotine+ are already mature, and the "free music" positioning easily slides into copyright gray areas. #Audio #Downloaders #Networking

A free music player based on the Soulseek network.

ns2250225Rust★10
B nuyoah-portrait-character-designer Breaks a portrait's vibe down into face shape and makeup, then turns it into Chinese prompts—essentially a structured prompt template. It saves some effort for character designers, but such tools are everywhere, and 46 stars in a day came too fast. #Prompting #Image Generation #AI

Nanyuan · Portrait Character Designer: turns a person's aura into facial structure, makeup and a complete Chinese prompt.

nuyoah-ai-worksPython★46
B venue-studio A browser-based venue seating chart editor with built-in venue templates and clearly defined scope; but its target users are event and venue operators, so generality is weak, and it doesn't explain export formats or collaboration. #Web Frontend #UI Design #Editors

Professional browser-based venue seating map editor with built-in arena layouts

topninjaCSS★9
B WMRL A paper's companion project page and code arguing for scaling automated research agents with world models, a relatively new direction; but beyond the paper there are no reproducible details, so for now it's just reading material. #ML Research #AI Agents #LLMs

Scaling Automatic Research Agents via World Models (arXiv:2608.12564) — project page and code release

xiyuanyang45Python★11
B deepseek-v41-flash-spark Runs MoE inference on a single DGX Spark using resident hot experts plus NVMe streaming loads, an idea with merit. But the repo calls itself WIP, the model name does not match any known release, and another repo on the same topic appeared the same day, so it cannot be verified for now. #LLMs #GPU #Performance #Hardware

DeepSeek-V4.1-Flash on a single DGX Spark (GB10): resident hot experts + NVMe streaming, DSpark, OpenAI API. Work in progress.

0xBakeerPython★49
B life-recorder From the browser-use team with nearly 180 stars in a day, but the repo has no description, so the name only suggests an app that records personal operations. The key issue for this topic is precisely privacy boundaries and storage, which are currently impossible to judge. #Privacy #Automation #macOS

browser-useSwift★176
B Travel-Plan-Page Turns travel plans into a nice-looking web page on your phone; the need is real but the bar is low, and similar static itinerary page tools are plentiful. Unless sharing, collaboration, or offline support is added, it's just a personal utility. #Web Frontend #Productivity #Documents

Quickly turn your travel plan into a travel webpage suitable for viewing and sharing on mobile.

do-tongxueJavaScript★13

COrdinary64

C gpu-time A small neural model for parsing dates and times in natural language, aiming to replace rule-based parsers like dateutil on ambiguous expressions. 55 stars in 0.7 days shows real niche demand for fuzzy time parsing in the agent era; the path is an embeddable library or agent tool that becomes a default dependency. The main risk is that rule-based approaches are already good enough, so adding a model only increases size and uncertainty. #ML Research #AI

A small neural model for parsing date & time

arikchakmaPython★55
C Main A concept repo that stitches a full fruit fly connectome simulation together with Coinbase AgentKit trading actions, seemingly riding both the connectome and AI trading hype at once. 237 stars in 0.5 days is topic-driven rather than real usage. The theoretical path is a neuromorphic simulation platform or strategy research tool; the biggest risk is unknown language and no verifiable output, a marketing shell under scientific packaging. #AI Agents #Crypto #Biotech

A full retained fly-connectome simulation with experimental memory and guarded Coinbase AgentKit trading actions.

flybook-gitLanguage unknown★237
C DeepSeek-V4.1-Flash-vLLM-DGX-Spark A deployment recipe for running a 552B MoE with vLLM TP4 on four DGX Spark machines, including sm121 kernel compilation, on-disk Engram patches and measured numbers, directly useful to teams building their own LLM inference. 51 stars in 1.7 days is precise but niche. Monetization is hard, at most deployment consulting; the biggest risk is that the model name and measurements cannot be verified and the recipe quickly goes stale with vLLM versions. #LLMs #GPU #Performance #Benchmarks

DeepSeek-V4.1-Flash (552B MoE, MXFP4 experts, 1M ctx) on four NVIDIA DGX Sparks with vLLM TP4: Engram-on-disk patch, sm121 kernel build, launchers, measured numbers

tonyd2wildPython★51
C WorldinWorld The official implementation of WorldinWorld, aimed at research reproduction in world models and embodied intelligence, saving peers the cost of rewriting training and evaluation code. 51 stars in 1.8 days come mainly from the lab's brand and topic hype rather than engineering adoption. Monetization can only come from later paper influence or enterprise collaboration, hard to charge directly; the risk is the maintenance gaps common to research code and a steady stream of similar implementations. #Video #ML Research #GPU

Official implementation of WorldinWorld

Westlake-AGI-LabPython★51
C rune It bills itself as a development environment for professional developers, written in Go, but the description is minimal and the positioning vague, unclear whether it replaces VS Code or is a CLI toolchain. 176 stars in 1.5 days shows strong topic appeal, but authenticity is unverified. The commercialization path is unclear, at best copying Cursor or Warp subscriptions; the biggest risk is that the dev environment track is locked down by giants and well-funded teams. #Editors #Terminal #GPU #AI Agents

the development environment for pros

unstablebuildGo★176
C Mac-Duo It ports the iPhone Duo folding blur effect onto the MacBook lid angle, a pure visual toy that solves no engineering problem. 523 stars in 1.2 days rides aesthetic hype, but six or seven clones like MacDuo, Bendable and hinge appeared the same period, showing zero barrier. Monetization is only a one-time paid utility and paid competitors already exist; the risk is clones driving pricing to zero and abandonment once hype passes. #macOS #iOS #Animation

Wish you could bring the iPhone Duo effect to your MacBook?

sumimakitoSwift★523
C youtube-scout Packages YouTube topic research as a Claude Code skill: input a topic, output a multi-sheet workbook sorted by views and engagement. It saves effort for ad research people, but it is essentially API scraping plus spreadsheets, similar scripts are plentiful, and most of the value lies in the prompt. #Agent Skills #LLMs #Data Analysis #APIs

Claude Code skill: one topic in, one ranked YouTube research workbook out. Searches YouTube Data API v3, ranks by views, engagement, momentum or breakout, and writes Videos, Channels, Summary, Comment…

AgriciDanielPython★9
C openmuse A personal assistant built on the company's own OpenComputer Serverless Agents, self-hostable, but the description stops at positioning: which services it connects to, where data is stored, and how to start are all unstated. More a platform example than a directly usable tool. #AI Agents #Self-Hosted #LLMs

A personal assistant you deploy, built on OpenComputer Serverless Agents

diggerhqTypeScript★29
C baron Local cross-session memory for coding agents, attaching a judgment to each answer—it targets the real pain of agents forgetting things. But the memory-layer space is crowded, and at public alpha it's unclear how it beats existing options. #AI Agents #LLMs #MCP

Baron Munchausen — local memory for coding agents that outlives the chat, with a verdict on every answer. Public alpha.

shinegangPython★11
C EnhancedFly A flight plugin for Mohist/Spigot 1.20.1, clear in one sentence, but the MC ecosystem is saturated with flight and permission plugins, with no visible difference from vanilla or similar plugins. 18 stars suggest server owners haven't noticed it yet. #Games #Plugins #minecraft

A flight plugin for Mohist/Spigot 1.20.1

YouDaoRSJava★18
C mac3 A native Metal build of GTA III on Apple Silicon that ships downloads but no source, effectively repackaging re3's work. Convenient for people who just want to play, but of no reference value to developers and in a copyright gray area. #Game Porting #macOS #Games #gta

Native Metal build of GTA III for Apple Silicon Macs — download only

gtamacShell★16
C meme-radar A local read-only multi-chain meme candidate scanner with manual review, helping traders filter on-chain noise into a candidate list. 77 stars in 0.5 days comes from sustained meme trading interest, but this track is crowded with highly similar tools. Monetization could be signal subscriptions or trading rebates, with gray-area compliance; the risk is heavy reliance on on-chain data sources and pressure from mature platforms like GMGN. #Crypto #Monitoring #Data Analysis

Meme Radar open-source edition: local read-only, multi-chain Meme candidate scanning and manual review tool.

nhovongoc0-maxJavaScript★77
C RTX-FG-Manager Manages an RTX frame-generation toggle via binary releases plus docs, with no source or implementation notes, so security and compatibility can't be verified; mature alternatives for GPU tuning already exist. #GPU #Windows #Games

RTX Frame Generation Manager by 小南瓜 · Windows binary releases and documentation.

pandaligxLanguage unknown★17
C MacDuo Native Swift plus Metal makes the desktop follow the lid angle, another implementation of the MacBook folding effect, slightly more technically complete than the batch of clones. 134 stars in 1.3 days mostly ride the topic rather than product differentiation. It can only be a one-time paid small tool with no repeat purchases; the risk is six or seven similar repos already existing, no first-mover advantage, and no maintenance after the hype fades. #macOS #Hardware #UI Design

A native Swift + Metal Mac app that makes your desktop follow your MacBook lid.

DhananjayBhosaleSwift★134
C macVC Same origin as the author's GTA III build, providing a Metal executable of Vice City for Apple Silicon with no source or build instructions. It's distribution rather than engineering contribution, and you still need to supply the game assets. #Game Porting #macOS #Games #gta

Native Metal build of GTA Vice City for Apple Silicon Macs — download only

gtamacShell★12
C StandUp A small Swift app that reminds you to stand up; the function is clear, but similar tools are plentiful in app stores and it lacks differentiation such as posture detection or habit tracking. Fine for personal use, not worth following as a project. #Health #macOS #Productivity

A reminder tool that prevents sedentary behavior and provides regular cycle reminders, developed natively based on Swift.

jiayouzlSwift★11
C Phantasm Fine-tunes chat logs into a local chatbot that imitates a person—straightforward, and digital-twin projects are already numerous. Privacy and ethical boundaries aren't discussed, nor are the data volume needed or the results, so practical value is doubtful. #LLMs #AI #Privacy

Turn someone’s chat history into a chatbot that talks like them. Extract the personality, fine-tune it, and run the result locally.

rugbedbuggPython★11
C DFLegacy A local mock server, launcher, and patches for some server's ACT1 client—private-server and reverse-engineering territory with obvious legal risk. Useful reference for people studying game protocols, but not something to recommend publicly. #Games #Reverse Engineering #Networking

A local mock server, launcher and client patch for the Seris server public beta ACT1 client.

shizuku30C#★16
C county-value-atlas A 3D interactive atlas of Zillow county-level home prices, visually somewhat interesting, but it's only data display with no reusable tool or method; the author shipped several homogeneous sites in one day, making it feel more like a portfolio. #Data Analysis #3D #Web Frontend

Interactive 3D atlas of upper-tier Zillow home values across U.S. counties

topninjaJavaScript★9
C groundtruth-atlas An atlas combining county-level rent capitalization rates with homeowner insurance risk, a finer topic than the price charts; but it's still static data display and highly homogeneous with the author's two other repos. #Data Analysis #Fintech #Web Frontend

Unified county-level atlas for housing cap rates and homeowners-insurance risk

topninjaHTML★9
C O.status A minimalist Android status indicator showing cellular, Wi-Fi, and battery. The implementation looks clean, but the system status bar already shows this information, so its value depends on specific models' beautification needs. #Android #UI Design #Productivity

A clean and minimal Android status indicator for cellular, Wi-Fi and battery status.

CATCHINGLKotlin★32
C VioletToolBox A multi-function toolbox for Android tinkering that does a bit of everything without focus, and the description doesn't clarify features or safety boundaries. Fast star growth reflects tinkerer-community spread, not quality. #Android #Automation #Productivity

A multifunctional device-tinkering toolbox for Android development.

Smart-PaocaiC#★76
C Docky.rs A dock/status bar for Hyprland written in Rust, a need already covered by waybar and hyprpanel. The repo is one sentence with no screenshots or configuration docs, making it hard to justify replacing existing setups. #Linux #UI Design #Productivity #hyprland

A Rust dock/bar for Hyprland

Volcha-ZRust★26
C wedding-video-guided-wizard A skill pipeline that splits wedding videos into 14 steps, chaining writing, voiceover, image generation, and editing; the steps are honestly written but it's essentially prompt orchestration with variable results, dependent on multiple external model accounts, and very niche. #Prompting #Video Editing #Agent Skills

A 14-step guided Skill for real-story wedding videos: story gathering, Kimi writing package, voiceover, external GPT image generation, video packaging, music and subtitle editing.

aaronyi97Python★13
C bce-tool-rs A Rust rewrite of ACE's MCP client, with the author honestly noting it's a fork of ace-tool-rs; the description doesn't say what was improved, features overlap with upstream, and ten stars in a day shows no sign of actual use. #MCP #LLMs #CLI

bce-tool: MCP client for Better Context Engine — codebase indexing, semantic search and prompt enhancement (fork of ace-tool-rs)

linqiu919Rust★10
C RCWS_v5 A public release of 3D models and a mechanical design manual for a remote weapon station, a rare subject; but it doesn't explain the design's origin, purpose, or safety boundaries, reading more like a personal design archive that's hard to reuse in engineering. #Hardware #3D #Robotics

Step file of RCWS v5 and mechanical design handbook

wildcard1719Language unknown★25
C wloc A rule module for Apple WLOC location rewriting on Surge, Shadowrocket, and Loon, essentially a config excerpt that iOS users doing location testing or privacy isolation can install and use. Similar modules are already common, the code is tiny, and it lacks boundary and risk notes, so it hardly fills ten minutes of reading. #Networking #iOS #Privacy

Apple WLOC network positioning coordinate modification module · Surge / Shadowrocket / Loon.

zxishereHTML★15
C FakePlayer-CE A community-maintained continuation of the Minecraft FakePlayer plugin that solves the upstream abandonment, a real need for server owners. But it's a handoff rather than something new, and it doesn't say which server versions are supported. #Games #Plugins #Self-Hosted

Community-maintained continuation version of the FakePlayer plugin

EndlessPixel-StudioJava★13
C PrimeBox A one-liner about 'running RB on devices like the Prime Go' that neither explains what RB means nor clarifies supported scope and flashing risks. Possibly useful to DJ hardware modders, but the current description isn't enough to act on. #Embedded #Hardware #Audio

Run RB on Prime Go and others

erhan-C★14
C pocketvla Claims to train a 3M-parameter VLA from scratch—an interesting direction, but it's just a slogan with no tasks, data, or reproduction path, so there's no way to tell if it actually runs. #Robotics #ML Research #AI

Building a 3M VLA from Absolute Zero

penpenzhPython★19
C awesome-reasoning-generalization A paper list on LLM inference generalization, giving researchers a quick entry point to locate directions; it is manually curated rather than engineering output. 139 stars in 1.9 days comes more from the topic's heat than maintenance quality. There is almost no commercialization path beyond a survey or benchmark; the risk is severe homogenization among list repos, high maintenance cost with low return, and easy replacement by more active peers. #LLMs #ML Research #Learning

Awesome paper list for reasoning generalization in LLMs.

tue09Python★139
C riskops An IT and cybersecurity risk register for Turkish enterprises: 5×5 assessment, action tracking, and reports, built with PHP and MariaDB without a framework. The need is real but the form is traditional, the audience is local internal control teams, and the open source community has no use for it. #App Security #Network Security #Self-Hosted

BT ve siber güvenlik risk yönetimi platformu — risk envanteri, 5×5 değerlendirme, aksiyon takibi ve kurumsal raporlama. PHP 8.3 + MariaDB, framework ve CDN kullanmadan.

CodeByPinarPHP★8
C GLiFormer A general multi-task Transformer encoder released by Knowledgator (the GLiNER authors), a credible direction. But the repo is one sentence with no weights, usage, or evaluation, so there's nothing to learn yet; wait for the model release. #ML Research #LLMs #Science

Generalist Multitask Transformer Encoders

KnowledgatorPython★11
C FrostFold A tech demo using Metal layerEffect to turn the UI into frosted glass as you tilt an iPhone; visually interesting, but no reusable component or API was extracted. Good as inspiration, not as a dependency. #iOS #UI Design #Animation

Tilt your iPhone and the UI stays put in space. A Metal layerEffect turns the screen into frosted glass.

askmaddyySwift★11
C dsh-deepseek-web-login Wraps the chat.deepseek.com web login session as a local LLM provider, including login capture and PoW solving. Technically interesting, but it bypasses the official API and terms and could be blocked at any time, so it's unsuitable as a dependency. #LLMs #Reverse Engineering #Plugins

Unofficial DSH (DeepSeek Harness) plugin: use chat.deepseek.com web models as an LLM provider - browser-login capture, PoW solving, SSE streaming, prompting-based tool calls.

cv-superdingJavaScript★9
C DeepSeek-V4.1-Flash-EXL3-vLLM-2x-DGX-Spark A DeepSeek quantized inference config for two DGX Sparks—saves trial and error for owners of the same hardware. But it's just an environment recipe with a very high hardware bar, and it's basically meaningless on other machines. #LLMs #GPU #Performance

DeepSeek-V4.1-Flash EXL3 vLLM recipe for 2x DGX Spark (GB10)

sfxnzPython★12
C dsv41-flash-4x-dgx-spark-ja A Japanese-language runbook for quantized deployment on four DGX Sparks, with measured data—more honest than similar config sheets. But it's likewise only meaningful to people who own this hardware; other readers won't get far. #LLMs #GPU #Performance

DeepSeek-V4.1-Flash on 4x DGX Spark (switchless ring, vLLM TP4, Engram on disk) - 日本語の引き継ぎ RUNBOOK と実測.

tenaiaiaiPython★10
C stampede A read-only replay of wallet rotation samples on Robinhood Chain, with terminal and web views, viewable without an API key. Somewhat interesting for on-chain behavior observers, but the audience is very narrow and wallet-tracking dashboards are already plentiful. #Crypto #Blockchain #Data Analysis #CLI

Read-only explorer of observed wallet rotations on Robinhood Chain (PONS v2): terminal + web, replay of a recorded sample without API keys

Argona7Python★21
C selfhosted A set of Shell scripts for building a self-hosted VPN that saves some configuration time, but it doesn't say which protocol or what threat model, leaving no distinction from the many one-click scripts online. #Network Security #Self-Hosted #DevOps

Repo for running own VPN servers

frkn-devShell★12
C codebuddy2api Wraps a CodeBuddy subscription as a local OpenAI-compatible endpoint—the purpose is clear: turn paid quota into a general API for other programs to call, skirting the terms of service. Neither stability nor ban risk is addressed. #APIs #LLMs #Self-Hosted

Use your CodeBuddy subscription as local OpenAI APIs.

maiphucgiangPython★24
C N64DoomRPGPort Ports the mobile Doom RPG to the N64, a niche and interesting subject. But the repo has only one sentence, with no build instructions, screenshots, or playable flow, so opening it now only lets you read code; better to look at existing Doom RPG reverse engineering. #Games #Game Porting #Reverse Engineering

The Doom RPG port for N64

BigChat88C★9
C ui Yet another shadcn/ui-based AI product UI kit, claiming single-file and 38 interaction modes. Such component libraries are heavily homogenized and stars are few; better to just use the official shadcn examples. #Web Frontend #UI Design #AI

UI by Halaska: a single-file React UI kit for AI products, built on shadcn/ui. 38 UX patterns and ~100 components.

Halaska-StudioJavaScript★10
C mistok-chytalnia Turns experience from searching Ukrainian village archives into a Claude Code skill plus methodology material, a very specific domain. But the deliverables are mainly prompts and notes, a content repo rather than a tool usable today. #Agent Skills #Knowledge Base #Learning

Skills, methods and raw materials for archival research on Ukrainian villages. Developed for "Mistok", released for everyone / Claude Code skills and methods for archival research on Ukrainian village…

OuFinxLanguage unknown★14
C red-sand-rally A 3D combat racing game built with Three.js in the browser, playable directly but more like a personal game demo, with no level design or multiplayer documentation. Useful as a reference for learning Three.js game structure. #Games #Game Dev #Web Frontend #3D

Red Sand Rally — a playable 3D combat racing game in the browser (TypeScript + Three.js).

Tao716TypeScript★9
C helloworld-bruteforced A joke repo that writes Hello World in the most absurd way possible; it has spread as a C language curiosity but is useless for any engineering problem. Stars come from amusement rather than value, making it a personal lark. #Learning #Debugging #Reverse Engineering

One of the most cursed ways of writing Hello World program in C

anhol0C★29
C Liduo A little macOS toy that bends, blurs, and dims the desktop when you close the lid. Fun as a Metal effect demo, but it solves no real problem and offers no reusable value. #macOS #Animation #UI Design

A macOS app: the desktop bends, blurs and darkens when the MacBook lid is closed.

cypresskirSwift★14
C fievel A Linux app that moves the pointer by keyboard instead of mouse, aiming to reduce mouse use. warpd, keynav, and others already do this, and the repo explains neither how it differs nor how to install it. #Linux #Accessibility #Productivity

A linux application that aims to replace mouse movements with keyboard!

MontyTheSoftwareEngineerRust★16
C DayVault A daily journaling app with an achievement system for iOS, optionally with AI companionship, reading like a personal practice project. Journal apps are plentiful, and the repo doesn't explain what sets it apart. #iOS #Productivity #AI

Daily records, achievements and an optional AI companion | Native iOS daily recorder with earned achievements and an optional AI companion.

Jack-Li-NpuSwift★8
C earth-moon-solar An interactive solar system demo built with Three.js and NASA textures; a common teaching demo with clean code, but it raises no new questions and extracts no reusable components. #Web Frontend #3D #Learning

Explore Earth, the Moon, and all eight planets in an interactive Three.js solar system with NASA imagery and JPL orbital data.

ethanplusaiJavaScript★9
C transformer-architecture Shows Transformer architecture through 3D interaction, but hangs a nonexistent model version on it as a gimmick; there are already several more solid interactive tutorials explaining the principles, so this looks more like a visual demo. #ML Research #3D #Learning

Explore the original Transformer and DeepSeek V4.1 Flash in interactive 3D.

petergptJavaScript★21
C HyperModifier A Magisk module adding a soft glass interface to HyperOS 4, a small theme beautification item. The audience is limited to Xiaomi flashing users, the description is too short, and it never covers compatible models or flashing risks. #Android #UI Design #Plugins

HyperGlassify soft-light glass completion; HyperOS4 system UI light beautification module.

AritxOnlyKotlin★10
C DeepSeek-v4.1-Flash-DGX-Sparks Attaches an unreleased DeepSeek model name to a multi-DGX Spark setup, with only one sentence of description and no scripts or measured data. Another same-topic repo appeared the same day, nearly a twin, looking more like keyword riding. #LLMs #GPU #Hardware #Performance

DeepSeek-V4.1-Flash on 3-4x NVIDIA DGX Sparks

MiaAI-LabPython★49
C iphone-duo-skill Organizes Apple Tech Talks and HIG into a Claude Code skill to help apps adapt to iPhone dual-screen form factors. The organization is tidy, but it's still a document summary, and several similar skills have popped up in the past couple of days. #Agent Skills #iOS #Learning

A Claude Code skill for making apps compatible with iPhone Duo. Organized based on 6 Apple Tech Talks and the HIG.

d-dateLanguage unknown★16
C airspace A 'the database you already have' tagline plus stars driven by the author's fame, but it doesn't clarify whether it's a query layer, cache, or sync solution. Vision first, usage absent; it can't be evaluated right now. #Databases #DevOps #APIs

The database you already have.

danielroeTypeScript★9
C ComfyUI-YuE2 A re-release of the YuE2 music generation node to keep up with newer ComfyUI and Transformers, a handoff fix. Some people do need it, but two similar re-releases appeared the same day, making it hard to tell which is canonical. #Audio #Plugins #AI

Dependency-safe compatibility fork of piscesbody/ComfyUI-YuE2 for modern ComfyUI and Transformers.

ScryptHunterPython★16
C ai-security-tool The description is stuffed with emoji and 12+ autonomous AI agents, cramming SSH, RDP, SBOM, and CTF into one HTML repo, so it's unclear what's actually finished. Security tools should avoid exaggeration; start with a reproducible minimal demo. #App Security #AI Agents #Network Security

🛡️ Free open-source AI-powered security terminal & vulnerability scanner (CVE, SBOM). Supports SSH, SFTP, RDP, VNC, Serial, and 12+ autonomous AI agents (DeepSeek, OpenAI) for automated security workf…

ZeroDayEvilHTML★14
C MangaWeave The direction of making image models think in comic pages rather than illustration panels is interesting, but the repo has no language info and no runnable entry point, so it's unclear whether it's a paper, dataset, or prompts. Panel consistency is a real problem, but reproducible results are needed first. #Image Generation #AI #Multimodal

Teaching image models to think in manga pages, not illustration grids.

anhe2021212-specLanguage unknown★9
C openai-image-prompt-writer Organizes official image-generation prompt advice into a Codex skill; it's template repackaging whose benefit depends mainly on the model itself. Similar prompt skills have appeared in bulk lately, making differentiation hard to see. #Prompting #Image Generation #Agent Skills

Codex skill for GPT Image 2.5 prompt writing, editing and reference-image workflows. Based on official OpenAI guidance.

gnipbaoPython★10
C hop-out An 'exit liquidity' monitoring terminal for tokens on a new chain, aimed at a tiny speculative audience with jokey marketing copy. Its lifespan will likely end along with that chain. #Crypto #Blockchain #Fintech

Your bag grew. The exit didn't — a read-only exit-liquidity terminal for Pons V2 tokens on Robinhood Chain.

insomnia-vipTypeScript★10
C x9-gpt A security POC claiming to extract memory data from chatbots using Duckscript, but the description is vague, with no attack surface, reproduction steps, or applicable models, so 'stealing all chatbot memory' reads like clickbait. LLM security researchers might glance at it, but it's currently scattered experiment notes rather than a usable tool. #LLMs #Jailbreak #Prompting #Privacy

Duckscript for LLM data extraction, stealing all chatbot memory. (POC for education purposes)

zr0nLanguage unknown★9
C deepseek-v41-flash-mac-mini Claims to run 'DeepSeek V4.1 Flash FP4 weights' on a 16GB M1, but both the model name and quantization scheme make no sense, and benchmarks were posted less than a day after the repo was created, looking more like hype-chasing. Bounded streaming loading in MLX is a real problem, but credible reproduction logs are needed. #LLMs #ML Research #macOS

Run original DeepSeek V4.1 Flash FP4/FP8 weights on a 16 GB M1 Mac mini with bounded SSD streaming in MLX. Recipe, benchmarks, and demo.

atbenderPython★14
C unifolm-wla A repo from the Unitree direction, but with no description at all, so it's unclear whether it's model weights, training code, or docs; it can't be evaluated until a README is added. #Robotics #ML Research #AI

unitreeroboticsLanguage unknown★15
C RP-Toolbox A free tool collection for the Retroid Pocket Nova and RP6 handhelds, described in a single sentence that reveals neither specific features nor installation, aimed at a very narrow audience. #Android #Hardware #Productivity

A free utility app for the Retroid Pocket Nova and RP6.

GuanRen-PanghuLanguage unknown★12
C typescript-engineering A TypeScript engineering 'skill pack' for AI coding assistants, essentially a collection of specs and prompts; such content repos have appeared in bulk lately and are hard to judge for quality. For actually learning type safety, the TS handbook and typescript-eslint docs are more reliable. #Prompting #Agent Skills #Code Quality

Reusable TypeScript engineering skill for AI coding assistants. Practical guidance on type safety, runtime validation, architecture, security, and testing. Works with Codex, Claude Code, Copilot, Curs…

bluenzzzJavaScript★8

DSkip60

D hinge Another Swift implementation that makes the MacBook desktop follow lid angle, nearly identical in function to Mac-Duo and Bendable. Its 70 stars suggest these folding-effect repos ride a shared topic rather than any single product; there is no business model, and free clones have already erased paid space, with the risk that it is pure eye candy whose novelty fades in a day or two. #macOS #Hardware #UI Design

Your MacBook desktop follows your lid.

NoveumSwift★70
D Bendable A macOS menu bar utility that drives an animation from the MacBook lid angle at screen refresh rate, part of the folding-effect trend with a hardware-gimmick bent. 78 stars in 1.1 days rides purely on topic hype with limited practical value. Commercialization is one-time payment only and already surrounded by free clones; the risk is reliance on non-public capabilities like the lid angle sensor, which a system update could break. #macOS #Animation #Hardware

macOS menu bar app that turns MacBook lid movement into a screen animation, driven by the lid angle sensor at display rate.

opensourcevillainSwift★78
D Trend-Rebalance-Map-Herman- By name it is a trend rebalancing indicator for TradingView, aimed at retail traders judging positions and trends, but the repo has no language tag and no README. 61 stars in 1.3 days look more like trading community funneling than code quality. Paid indicators or community subscriptions could monetize it, a typical gray area; the risk is a flood of similar indicators, no backtest validation, and easy descent into a marketing shell. #Fintech #Crypto #Data Analysis

Trend Rebalance Map [Herman]

HermanTradingLanguage unknown★61
D flycoinrh A concept that ties the real 160,000-neuron fruit fly connectome to Robinhood on-chain token issuance, assembling the connectome and crypto hype together. 116 stars in 1.3 days looks more like speculative traffic than research adoption. Theoretical monetization is a neural simulation platform or token narrative, but scientific credibility is doubtful; the biggest risk is abandonment once crypto sentiment fades, so it cannot serve as a technical dependency. #Crypto #Biotech #AI

A real fruit fly connectome, 165,122 neurons, launching tokens on Robinhood Chain

fruitflydevPython★116
D MacBook-Duo It treats lid angle as a dual-screen interaction controller, part of an obvious clone wave alongside Mac-Duo and Foldelight. Its 67 stars in 0.8 days are topic spillover with minimal technical difference; it can only be a one-off paid utility with little repeat use, and the main risk is zero moat, since a change in sensor permissions or APIs breaks it and nobody maintains it after the hype. #macOS #Animation #Hardware

A macOS utility that uses the MacBook lid angle as an interactive dual-screen control.

jlxc2001Swift★67
D iphone-duo-macos-animation It recreates the iPhone Duo folding progressive blur animation on MacBook using native Swift, one of the earlier clones in this lid-angle wave. Its 51 stars in 1.5 days are entirely topic-driven, and technically it only reads the tilt sensor; there is no commercialization space, already drowned by many free implementations, and the risk is pure visual effect with very low retention. #macOS #Animation #Hardware

A clone of Iphone Duo's fold progressive blur animation on macbook using lid tilt sensor. Fully native written in swift! Demo:

lqSky7Swift★51
D fivem-visual-enhancer It claims to be a FiveM visual overlay, player tracker and aim assist, but is actually a typical free cheat distribution page funneling traffic by requiring no key. 50 stars in 0.4 days come from cheating demand and suspected star inflation, not real momentum. There is no legitimate commercialization, only bundled malware or later paid unlocks; the risk is bans and legal liability, and the repo itself is untrustworthy. #Game Cheats #Anti-Cheat #Scams

FiveM External Utility 2026 | Visual Overlay, Player Tracker, Aim Assistant | Free, No key required

InfernalgrunozzleLanguage unknown★50
D dota2-overlay-toolkit A Dota 2 cheat for map vision, auto-combos and last-hit tracking, with a description format almost identical to the same day's FiveM and PUBG repos, suggesting mass production from one template. 50 stars concentrated in 0.4 days look more like star inflation than organic growth. The monetization playbook is free traffic then a premium version or trojans; the risk is very high: anti-cheat bans, legal risk, and code of unknown origin. #Game Cheats #Anti-Cheat #Scams

Dota 2 External Overlay Utility 2026 | Map Vision, Auto-Combo, Creep Tracker | Free, No key required

PentagonTurtleLanguage unknown★50
D pubg-visual-toolkit A PUBG player radar, tracking and aim assist tool whose description format exactly matches the same batch of Dota2 and FiveM repos, marking it as pipeline-produced cheat ware. Its 50 stars represent no real momentum or technical substance; the business is gray-market distribution and secondary charges, with risks of permanent anti-cheat bans and likely embedded credential-stealing malware. #Game Cheats #Anti-Cheat #Scams

PUBG Visual Toolkit 2026 | Player Tracker, Radar Overlay, Aim Assistant | Free, No key required

SwordsmithCanyonLanguage unknown★50
D virtual-library-guide A build guide fed to platforms like Lovable and Replit, teaching models to help you build a personal book library. It's tutorial and prompt content whose value depends on the author's own implementation, and the repo contains no runnable finished product. #Prompting #Learning #Knowledge Base

a guide you can give your vibe coding platform of choice (lovable, replit, claude code, etc). to walk you through how to build my virtual library

carollia99Language unknown★10
D dsh-bg-new A sidebar plugin that changes the background of some internal system's web page—just a wallpaper drawer plus frosted glass. Small in scope, but the audience is basically the author alone, with little reusable value. #Web Frontend #Plugins #UI Design

DSH web background changer: the sidebar "Wallpaper" button opens a right-side drawer supporting system presets/solid colors/gradients/images/videos, with scroll wheel + thumbnail dragging controlling …

sz1698JavaScript★19
D awesome-gpt-image-2-5-prompts A prompt collection with example images, marked as officially maintained by a certain model, essentially a content repo; it's a reference for image generation but not a tool, and similar lists are already abundant. #Prompting #Image Generation #Learning

Curated GPT Image 2.5 prompts, visual examples, and source-backed creative workflows. Maintained by Image 2.5.

wangrunlinJavaScript★28
D awesome-kicad-projects A KiCad project list that's pure link collection with no reviews or categorization criteria; hardware fans might find inspiration, but its value depends entirely on the quality of others' repos. #Hardware #Learning #Embedded

📔 📚 A curated list of Project Made with Kicad.

way2pramilLanguage unknown★42
D equivalence-driven-learning-v0.4-search-v1 Self-described as a non-gradient structured learning research prototype, plus a search engine v1, with title and version numbers piled together and no reproducible experiments or baselines. Research-method projects are worst served by this kind of self-talk. #ML Research #Science #AI

Equivalence-Driven Learning v0.4: structural, non-gradient learning research prototype with Search Engine v1.

MMVFIRMPython★13
D mlxtop Only a repo name and not a single sentence of description; from the naming it's presumably an MLX monitoring tool. It doesn't even say what it does, and hitting 42 stars in a day is no verifiable basis—not worth opening. #CLI #Monitoring #macOS

maximpriRust★42
D Awesome-Robot-Use-Agent A resource list for bot-operating agents with no description or selection criteria, a typical awesome list; it adds little information, and you won't remember many entries after reading. #Learning #Robotics #AI Agents

visitworld123Language unknown★16
D code-collection-share Code snippet collection and sharing that is essentially a content repository, with no explanation of technical implementation or usage, offering limited value to developers. #Knowledge Base #Learning #Documents

Code snippet collection and sharing.

L0NE-6Python★11
D 3d-Camera-control-H3-Minimax Both title and description are extremely vague, seemingly a UI for some 3D camera control experiment, with no stated inputs, outputs, model dependencies, or run instructions, so it can't be evaluated. #3D #Image Generation #UI Design

Camera control UI 3d Minimax

NyckMPython★18
D cumcm2026 A collection of papers, code, and reproduction scripts for problems A/B/C of a math modeling competition—contest-entry material. Somewhat useful to people preparing for similar competitions, but it produces no reusable tools, and solutions to such problems appear every year. #Learning #Science #Data Analysis

2026 Higher Education Press Cup National College Student Mathematical Modeling Contest A/B/C three-problem collection: complete modeling, solving, multi-method verification and one-click reproduction …

li2396803Python★19
D team-Omzo A hackathon entry repo with no description, and 208 stars in one day looks more like teammates canvassing votes; it's unclear what was built, let alone whether it's usable. #Games #Game Dev #Web Frontend

xsolla-baku-gametech-hackathonTypeScript★208
D tessra Self-described as a version control system for the AI era, with only a one-line vision and no design docs, protocol description, or runnable code. This kind of manifesto repo is best skipped. #AI #Collaboration #CLI

The version control system for the era of AI.

NitrusAphalionRust★15
D Parasite-Eve-2-HD-Remaster An HD remaster repo with no description, only a game name as the title, so it's unclear whether it's a texture pack, patch, or repost. With 37 stars in under half a day, it looks more like a fan gathering than an engineering project worth a look. #Games #Game Porting #Images

faligameLanguage unknown★37
D OpenFluxAndroid Claims to be an Android wrapper for OpenFlux, but doesn't say what OpenFlux is, which layer is wrapped, or how to run it; with only one sentence, its value can't be assessed. #Android #Plugins #SDKs

OpenFlux wrapper for android devices

p1neappleXpressC★10
D belltoken An official announcement repo for a Base chain token, with contracts and docs and no reusable technical content. The stars likely come from the token community rather than developers, so it offers no information for list readers. #Crypto #Blockchain #Marketing

Official BellToken ($BLL) project — Base Mainnet token, documentation, smart contracts and project information.

BellTokenBLLSolidity★10
D orbiters A "base camp" repo cramming a CRM, community homepage, and marketing site into one repository, with no runnable product and no clear boundaries—a classic case of trying to do a bit of everything. #Collaboration #Marketing #Web Frontend

One repository for every Orbiters project: PigroCRM, the Orbiters hub and the public website.

joinorbitersPython★10
D ComfyUI-YuE2 A ComfyUI custom node, but with no description of what it does or how to pass parameters; it's unclear whether it's a model wrapper or a workflow, so it can't be used. #Image Generation #Plugins #comfyui

piscesbodyPython★9
D alpha-harness No description at all; from the name alone it's impossible to tell whether it's a quant backtesting or evaluation framework. Without docs there's no usability—skip it. #ML Research #Benchmarks #Automation

residual-labPython★25
D team-XP-Farmers Another team's repo from the same hackathon, also without a description, so there's no way to judge whether the code runs; it's a one-off competition exercise. #Games #Game Dev #godot

xsolla-baku-gametech-hackathonGDScript★14
D MHI2Q-CarPlay-RGI-MMI-Mirror No description; the name suggests scripts for hacking CarPlay and MMI screen mirroring on Audi MHI2Q head units, a specific car-modding niche. It documents neither risks nor steps, making reproduction hard for others. #Reverse Engineering #Hardware #Automation

Lanye-zShell★14
D instagram-reporting-bot A bot for mass-reporting Instagram accounts, with no description whatsoever. Such tools abuse platform reporting mechanisms and offer no accumulable technical value even to their authors, so they do not belong here. #Scams #Social #Automation

ConjurerColumnLanguage unknown★11
D pub-dsh-privacy-router No description, yet 48 stars arrived quickly with no visible usage; the name hints at privacy routing, but implementation reliability can't be verified, so it's not recommended for real traffic. #Privacy #Networking #Self-Hosted

LYiHubJavaScript★48
D Comfyui-YuE2-T8 An undescribed YuE2-related node, hard to distinguish from several same-direction repos that appeared the same day, so usability and source reliability can't be assessed. #Audio #Plugins #AI

T8marsPython★20
D dsh-prompt-optimizer Nothing beyond the repo name, so it's unclear what kind of prompt it optimizes, who it's for, or how to use it; 21 stars in a day lack verifiable usage signs. Revisit once the README explains its purpose. #Prompting #LLMs

WestFox-AwAJavaScript★21
D wx-persona-agent A WeChat persona agent, but the repo has no description or usage, so it's unclear whether it's prompt wrapping or real memory and character modeling. No information to evaluate—skip it. #AI Agents #Social #LLMs

miaomiao6606Python★21
D fruitless The repo description is empty, so its purpose and technical direction can't be judged; even if the code is somewhat interesting, a project with no explanation isn't worth prioritizing. #Productivity #Automation #CLI

nicodunksPython★21
D SNIPE The repo has no description at all, so it is unclear what it does or who it is for. It gained 29 stars in just over a day, yet there is no verifiable usage or screenshot, so by the rules it can only be skipped. #Automation #CLI #Web Frontend

0xwhrariJavaScript★29
D CosyRedactGateway No description at all; the name hints at something related to a desensitization gateway, but there is no telling what it does, how to use it, or what it depends on. Authenticity cannot be judged, so it can only be skipped. #Privacy #App Security #Self-Hosted

CassiopeiaCodeJavaScript★18
D YOINK No description; the name plus a curve of 51 stars in half a day looks more like riding a trend, and its purpose and implementation cannot be evaluated. Not recommended to open. #Automation #CLI #Web Frontend

DefiLeooPython★51
D ComfyUI-Krea2-Accel The description literally says a ComfyUI plugin or whatever, showing the author hasn't even decided what it does, so usability can't be judged. #Image Generation #Plugins #AI

ComfyUI plugins or other stuff.

Q0809Python★10
D weathermapreal No description; only the language and name suggest a weather map Android app, with no information available for evaluation. #Android #Data Analysis #APIs

Rnitin007Kotlin★14
D bank-sampah No README, and nearly a thousand stars in one day clearly don't match the project's size, suggesting star farming. With no visible purpose or deployment method, it scores low by the rules. #Self-Hosted #Web Frontend #Databases

FaizpiPHP★918
D pku_quantllm A course discussion repo containing no runnable code or teaching material, just a notice board for a quantitative finance class. It has no value to anyone outside the course. #Learning #Fintech

Discussion area for the course on principles and implementation of modern quantitative trading systems.

aslan9Language unknown★9
D facebook-reporting-bot A bot for mass-reporting Facebook accounts, with no documentation at all; it's a gray-market tool abusing platform reporting mechanisms, with no technical content to evaluate. #Scams #Social #Automation

MountainzumprotectLanguage unknown★11
D iphone-duo-motion-study No description at all; the name only hints at iPhone dual-screen and motion effects, and it's unclear whether the deliverable is code, research, or assets. Not worth opening until a README is added. #iOS #Animation #UI Design

bravohenryJavaScript★13
D iphone-duo-skill A Shell repo with no description whose name collides with another iPhone dual-screen skill from the same period; it's unclear whether it's scripts, templates, or pure documentation. Not enough information to judge. #iOS #Agent Skills

eisenjimmyShell★11
D Mrite-dev Marketed as a mathematical modeling agent, it is actually a cracked distribution with activation checks removed, openly begging for stars. It has no usable value for developers, the binaries are of unknown origin and carry security risk, and projects that bypass licensing do not belong on this list. #Crack #Activation Bypass #AI

Mrite Mathematical Modeling Agent 2.6.13 dev version, activation check removed, native features fully preserved, continued updates to follow, leave a star.

06xxlinJavaScript★9
D JammersSimulator-Tool No repository description, and the name points to some game utility, with neither usage nor source documentation, so its real purpose can't be confirmed; it scores low by the rules. #Games #Automation #CLI

Jammers-Simulator-LabPython★14
D BetterAgent No description; the name only suggests something agent-related, with no indication of what it does or how to integrate it, so quality can't be judged. #AI Agents #LLMs #Automation

LING71671Language unknown★10
D PTSP-2nd No description, and the name can't be matched to any specific project or purpose, so by the rules it isn't listed. #Games #Plugins #Automation

RustyDoctorLanguage unknown★8
D ai-agent-langgraph-main A LangGraph practice repo with no description or README, and a name that looks like a course assignment output; it's impossible to tell what problem it solves. #AI Agents #LLMs

charlie-channPython★21
D 2026Fall A repo with no description whose title looks like a semester or batch directory; it's neither a tool nor readable material, and the source of its forty-plus stars is unclear. #Learning #Science #Knowledge Base

hnamkuLanguage unknown★44
D 2anh-zalo-bot No description at all; from the name you can only guess it's a Zalo chatbot. What it does, who it's for, and how to run it are all unclear—impossible to evaluate, and not worth ten minutes. #Social #Automation #APIs

luonghaianh1208Python★9
D GPT-Policy-Eval An empty-description repo whose name only hints at evaluating GPT strategies or policies, with no data, methods, or conclusions to check. #LLMs #Benchmarks

cheng-hahaLanguage unknown★19
D technocore-sonnet-challenge An empty-description repo whose name looks like a temporary directory for some contest or experiment, with no readable description, results, or code entry point. #LLMs #Benchmarks

flop-labsPython★9
D fc-27-pc-install-performance-companion Titled as an FC 27 build and performance companion, but there's no code—just keywords piled into a description. Sibling repos appeared in a batch the same day: typical game-name keyword farming with no added information. #Games #Windows #Performance

EA SPORTS FC 27 PC companion for installation preparation, Steam and EA app setup, controller profiles, graphics optimization, laptop performance and gameplay tracking.

maxwellhunt-19Language unknown★10
D cyberSecurity_DSATM No description; from the name it looks like a cybersecurity course repo for some college, with no tools or materials documented, so it's worthless to anyone outside the course. #Learning #App Security

kabalacLanguage unknown★9
D warframe-cheat-2026-frame-weapon-mod-tool "Cheat" is right in the name: loadout and resource tracking are just a cover—it's really a cheat repo with nothing reusable for developers, and such repos often carry suspicious binaries, so don't open it. #Game Cheats #Games #Windows

Warframe Cheat 2026 themed PC companion for frame and weapon builds, mod configurations, Prime collection, resources, farming and progression tracking.

leocollins478Language unknown★20
D ea-fc-27-pc-setup-gameplay-hub Nearly identical to the previous entry, an EA FC 27 build hub with a vague description and no implementation—clearly a batch-generated keyword repo that offers nothing to players or developers. #Games #Windows #Performance

EA SPORTS FC 27 PC setup hub for installation planning, Windows and laptop configuration, graphics presets, controllers, gameplay profiles and performance tracking.

mboyd-1240Language unknown★10
D bg3-cheat-engine-character-build-toolkit The title openly says Cheat Engine character build tool, but it's really a funnel page for a Baldur's Gate 3 mod with no technical content; such repos also carry copyright and malicious-file risks. #Game Cheats #Games #Windows

Baldur's Gate 3 Cheat Engine themed PC toolkit for character builds, party planning, equipment, quests, campaign progress and BG3 gameplay tracking.

mjohnson3164Language unknown★11
D Star-Wars-Zero-Company-Trainer-2026-Download The title is literally "mod download," with no source or technical content—a typical cheat funnel repo that may also bundle suspicious binaries; not recommended for anyone to open. #Game Cheats #Games #Windows

Star Wars Zero Company Trainer 2026 for PC

samberg-1998i4Python★12

Show HN 2

Show HN: Toast, a by default in-terminal IDE

Show HN

A terminal-oriented out-of-the-box IDE that aims to replace Neovim's configuration hell with a complete setup that looks good by default, immediately usable for TUI users. 187 stars in 0.3 days and 72 points as an HN hot post make it one of the strongest momentum items in this batch. Commercialization is hard, limited to sponsorships or an enterprise edition; becoming a standard dependency depends on the plugin ecosystem. The main competitors are Helix, Zed and LazyVim.

https://paradise-runner.github.io/toast/

#Terminal #Editors #CLI

▲72

Show HN: Spanda – Sub-microsecond LLM epistemic uncertainty in Rust

Show HN

It does zero-cost epistemic uncertainty quantification and hallucination detection for LLM inference, with a Rust implementation claimed to be 90,000 times faster than Semantic Entropy, targeting the latency pain of production hallucination monitoring. But Show HN has only 12 points and the repo 17 stars, so momentum is near zero. The path is embedding into vLLM or inference gateways as an enterprise hallucination monitoring API; the biggest risk is that the speedup cannot be reproduced and probe-based approaches face fierce competition.

#LLMs #AI #ML Research

▲12

Hot New Repos

Faizpi/bank-sampah★918
sumimakito/Mac-Duo★523 Wish you could bring the iPhone Duo effect to your MacBook?
angusdevgo/IDM_Pro_Tool★504 IDM activation and status maintenance tool.
yureii1996/cek-probe-model★354 Untuk cek model apakah maskingan atau bukan
kevinzakka/mjbatch★333 A Python library for running thousands of MuJoCo simulations in parallel on CPU
yudaprasetya007/routeVSCODE★327 ⚡ Zero-reload dynamic AI model switcher & local proxy for VSCode Copilot Chat via 9Router
deepseek-ai/deepseek-recipe★301
nftechie/stonkfly★298 A full retained fly-connectome simulation with experimental memory and guarded Coinbase AgentKit trading actions.
flybook-git/Main★237 A full retained fly-connectome simulation with experimental memory and guarded Coinbase AgentKit trading actions.
yi1108/printfilm★216 PRINTFILM: an AI science video and comic drama creation platform

A · Worth watching

Show HN 2

Hot New Repos

Star gain 1d