09.05 Today's Picks

302 new projects found today, 149 rated.

Created 2026-09-03 – 2026-09-05Fetched 600Filtered 50Rated 40

A · Worth watching

FlashRec

sohu-mptcPython

A CUDA Graph inference engine for generative recommendation supporting wide beam search and FP8 serving, solving recommendation latency and throughput bottlenecks. 85 stars show AI infrastructure interest; a recommendation service or open standard could commercialize, but NVIDIA Merlin and in-house systems compete, and ecosystem lock-in is the risk.

FlashRec is a CUDA-graph engine for generative recommendation: wide beam search (3–5 SID steps, n=50–512+) over a trie-constrained catalog, in-process FP8 serving, and ranked beams on /v1/chat/complet…

#LLMs #GPU #Performance #ML Research

★85

vulnometry

san3ncrypt3dPython

Replaces abstract CVSS scores with a business exposure index integrating NVD, EPSS, KEV and other data, solving vulnerability prioritization. 55 stars show security practitioner approval; SaaS subscriptions could commercialize, but RiskSense and internal systems compete, and delayed data source updates are the risk.

Measure what a vulnerability is worth to your business, not how severe it is in the abstract. Business Exposure Index scoring over NVD, EPSS, CISA KEV, OSV and GHSA.

#App Security #Network Security #Data Analysis #Monitoring #nvd #epss

★55

nopasswd-sudo

sudoingXShell

Time-limited passwordless sudo for AI agents: a systemd timer revokes it on expiry, visudo validates before install, and sudoreplay keeps an audit trail. The scope is restrained and the description honest, filling the security gap around letting agents run system commands; the limitation is it only covers Linux plus systemd.

Time-boxed passwordless sudo for agents. Grants expire on a systemd timer, validated with visudo before install, audited with sudoreplay.

#CLI #Linux #AI Agents #Privacy

★15

usque-custom-pro

KJGX66FJavaScript

Visually generates Cloudflare WARP/MASQUE configs and converts them to Clash, sing-box and other clients, simplifying proxy setup. 173 stars in two days show strong demand from censorship-circumvention users; monetization could be paid config services, but Cloudflare terms and regional blocking apply, against the official WARP client.

A visual registration, config generation, and multi-client conversion tool based on Cloudflare WARP / MASQUE. Generation, Clash/Mihomo smart routing, Shadowrocket, sing-box, local VLESS bridging, and …

#Networking #CLI #Automation #Self-Hosted #cloudflare #warp

★173

psd2live

tsunehimatoiKotlin

Automatically converts layered PSDs into Live2D models with layer recognition, mesh subdivision, physics and animation, solving time-consuming manual Live2D rigging. 39 stars show VTuber community demand; a subscription tool could commercialize, but official Live2D Cubism competes, and unstable generation quality is the risk.

PSD2Live is an automated Live2D model generation pipeline and desktop application. Given a layered PSD file, the system automatically performs layer semantic recognition, connected-component bilateral…

#Animation #Images #3D #Productivity #live2d #psd

★39

halo-kmp

AndroidPoetKotlin

Brings iOS Live Activities and Android 16 live notifications under one Kotlin Multiplatform API—a genuinely missing off-the-shelf solution for cross-platform status bars. The boundaries are clear and it suits teams already on Compose Multiplatform; but it's one day old with no release artifacts or examples, so read the source before integrating.

Halo: Kotlin Multiplatform Live Activities. ActivityKit on iOS, Android 16 live-update notifications, one API

#Android #iOS #SDKs

★10

data-deletion-skill

0nicellLanguage unknown

Turns personal data deletion into a Claude skill: 328 data brokers, 13 privacy-law jurisdiction templates, and escalation paths for 21 regulators. People cleaning up their data get an actionable checklist rather than a privacy manifesto; the limit is that the list's accuracy can't be verified, and results depend on whether the other side cooperates.

A Claude skill for running structured personal-data erasure campaigns — 328 data brokers, OSINT tools and breach-search engines with verified privacy contacts, request templates for 13 privacy regimes…

#Privacy #Agent Skills #LLMs

★9

pronote-digest

yoanbernabeuTypeScript

Aggregates a Pronote iCal subscription feed daily into next-day classes and homework, pushed to parents and students by email or other channels, with both a GitHub Action and a CLI. The niche is small and clear, and French school users can use it right after installing; the limitation is tight binding to the Pronote ecosystem.

Chaque soir, le planning ou les devoirs du lendemain depuis les flux iCal Pronote, par mail ou autre canal. GitHub Action et CLI.

#CLI #Email #Productivity #pronote

★19

coffer

dahliaRust

A Rust client that natively reads Apple Passwords/iCloud Keychain on Linux, addressing the real pain of cross-platform password access with a restrained scope. The risk is reliance on Apple's private protocols, so stability and compliance need your own assessment.

Native Linux client for passwords and other credentials stored in Apple Passwords (formerly iCloud Keychain)

#Secrets #Linux #macOS #Privacy

★9

ATHER-OBD-READER

SAM0-0C++

Reads the CAN bus directly from an Ather electric scooter's BMS, exposing cell balancing and health data the official app doesn't provide. A genuinely useful small hardware project for owners and mechanics of this niche model; limits are it covers one vehicle and requires your own wiring.

A open-source OBD data reader for Ather electric scooters. This device reads real-time CAN bus data directly from the scooter's BMS (Battery Management System), allowing owners and technicians to moni…

#Embedded #Hardware #IoT

★12

DogLegs

YibinWuC++

Code accompanying an IROS 2025 paper, using multi-leg-mounted IMUs for legged-robot proprioceptive state estimation, with a concrete problem and a citable method. A rare open implementation for quadruped state-estimation researchers; limited by its specific sensor layout, hard to port short-term.

[IROS 2025] DogLegs: Robust Proprioceptive State Estimation for Legged Robots Using Multiple Leg-Mounted IMUs

#Robotics #ML Research #Science

★9

cc-readback

affirmitvTypeScript

A read-only MCP server that reads session logs from ~/.claude into Claude Desktop, with secret redaction. Immediately usable for anyone running both Claude Code and Desktop, and it stays within bounds; the risk is dependence on local file formats that break when upstream changes.

Read your Claude Code sessions from Claude Desktop. A local, read-only MCP server over ~/.claude with secret redaction.

#MCP #AI Agents #LLMs #Privacy

★16

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

BHas a spark31

B on-device-3d-scanner Generates Gaussian splatting 3D scans fully offline on iPhone from 8 photos in about 3 seconds, solving mobile 3D scanning's cloud dependency. 28 stars show AR creator interest; a consumer 3D scanning app could commercialize, but LiDAR scanning and Polycam compete, and accuracy and performance are the risks. #3D #iOS #ML Research #Multimodal

3D scanner running fully on-device: 8 photos to a spinnable gaussian-splat scan in ~3s on an iPhone, via Depth Anything 3 + react-native-executorch + TypeGPU

NorbertKlockiewiczTypeScript★28
B plain-prose An Agent skill that removes AI writing traces from English, Russian and German, solving AI-generated text detection. 66 stars show content creator demand; a writing plugin or API could commercialize, but Grammarly and AI detectors compete, and language coverage is limited. #AI #Agent Skills #Content Writing #Productivity

Agent skill that removes AI writing patterns from English, Russian and German prose. Merges stop-slop and avoid-ai-writing, adds a zero-dependency checker.

dripipsJavaScript★66
B WasmQuay Offline decodes WebAssembly components and classifies host capabilities, gating them by policy to solve WASM supply chain security. 36 stars in two days show security community interest; CI security scanning could commercialize, but wasm-tools and official tools compete and the ecosystem is immature. #App Security #Reverse Engineering #WebAssembly #CLI

Offline WebAssembly component inspection and capability-policy toolkit - decode .wasm binaries, classify host capabilities, gate against explicit policies. Rust + TypeScript, zero dependencies.

britsmanRust★36
B DLSS-NR-on-AMD Runs NVIDIA DLSS 5 neural rendering on AMD GPUs by reverse engineering around hardware limits, solving DLSS unavailability for Radeon users. 444 stars in two days show strong demand; commercialization is nearly zero as a gray hack, and the biggest risks are NVIDIA legal action or driver updates breaking it. #Reverse Engineering #GPU #Games #dlss #amd

Run DLSS 5 Neural Rendering on your AMD GPU. Magic!

danielblncLanguage unknown★444
B paddock A native inference server in Rust compatible with OpenAI and Anthropic APIs and multiple quantization formats, solving self-hosted model deployment. 35 stars show developer interest; enterprise private deployment could commercialize, but vLLM and Ollama compete, and ecosystem maturity is the risk. #AI #LLMs #APIs #Performance #openai

Native Rust inference server for open models on NVIDIA GPUs. OpenAI- and Anthropic-compatible APIs, GGUF + safetensors, FP8/NVFP4/MXFP4/Q8/Q4, built-in Studio

truesparRust★35
B my-bad Beautifies dev server error pages with live updates, solving hard-to-read error messages during debugging. 58 stars come from a known author, so momentum is credible; a framework plugin could commercialize, but Vite and Next.js built-in error overlays compete, and ecosystem lock-in is the risk. #Debugging #Web Frontend #CLI

Beautiful dev-server error pages that live update. ✨

danielroeTypeScript★58
B kugiri Splits text into lines, words, and characters at the browser's actual line-break positions, solving the long-standing guesswork in subtitles and per-character typesetting, doing one thing with an accurate description. The downside is narrow applicability; only people doing text layout will think of it. #Web Frontend #UI Design #Documents

Splits text into lines, words and characters exactly where the browser already broke it.

edoardolunardiTypeScript★11
B ps5_kernel_loader An IDA Pro 9.x loader for PS5 kernel dumps, ported from the PS4 tool of the same name, by a well-known scene figure who honestly notes Claude's involvement. The audience is extremely narrow but the thing is real; aside from console firmware researchers, basically no one can use it. #Reverse Engineering #Hardware #ps5 #ida-pro

ps5 kernel loader for ida pro 9.x . created by myself and claude, based on socratic bliss's ps4 one

zecoxaoPython★9
B Penelopa.ai Analyzes Codex and Claude Code sessions to extract repeated workflows and turn them into reusable skills, improving AI coding agent efficiency. 32 stars show team demand; an enterprise agent optimization platform could commercialize, but LangSmith and others compete, and data privacy is the risk. #AI Agents #Agent Skills #Productivity #Code Quality #codex #claude

Continuous improvement for AI coding agents: Penelopa analyzes real Codex and Claude Code sessions, finds repeated workflow patterns, and turns them into reusable skills, checks, prompts, and recommen…

chigwellTypeScript★32
B PromptLedger Provides version management, deterministic comparison snapshots and experiment reports for prompt templates, solving irreproducible prompt iteration. 42 stars show AI engineering demand; a team collaboration tool could commercialize, but LangSmith and Promptfoo compete, and ecosystem integration is the risk. #Prompting #LLMs #Testing #CLI

PromptLedger - version prompt templates, run deterministic comparison snapshots, score candidate responses, and ship diffable experiment reports. Python core plus a Go companion CLI.

cyptnpassiraPython★42
B CacheCraft Replays request traces to test different semantic caching strategies and quantify real gains, solving hard-to-evaluate LLM cache effectiveness. 36 stars show AI cost optimization demand; a cache strategy library or SaaS could commercialize, but Redis and LangChain caching compete, and benchmark representativeness is the risk. #AI #LLMs #Performance #Benchmarks

Semantic cache strategy bench: replay recorded request traces through candidate caching policies and measure which one actually pays off.

mpaymenremoraPython★36
B unpeel A Swift terminal app plus a Rust multiplexing server, aiming to keep AI agent sessions in the terminal alive without dropping. The direction is concrete and more substantial than a pile of empty agent-orchestration scaffolds; but the repo is under a day old, usage docs are thin, and it goes head-to-head with tmux and claude-squad-style solutions. #Terminal #AI Agents #CLI #macOS

A terminal app & rust multiplexer server for always-on terminal AI agents

unpeel-comSwift★17
B omarchy-workspace-layout Gives Hyprland per-workspace remembered tiling layout ratios, adjustable by dragging dividers, solving messy layouts across multiple monitors and workspaces. 99 stars show Linux user approval; commercialization is hard but it could become a standard Omarchy ecosystem plugin, with Hyprland interface changes as the risk. #Linux #UI Design #Productivity #Automation #hyprland #omarchy

Per-workspace tiling ratios for Hyprland, on Omarchy. Drag the dividers and your windows move — 25/50/25, master and stack, thirds, grid. Named layouts, per-workspace assignment, switchable profiles.

bjarneoJavaScript★99
B packslip A Rust tool that signs manifests for vendored binary releases, by the mise maintainer, similar in approach to aqua's verification manifests. Useful for supply-chain verification in build and CI scenarios; still a prototype with thin documentation. #Build #App Security #Secrets

A signed release manifest for vendor binaries

jdxRust★9
B adaptive-frame-extractor A C++ tool for adaptive frame extraction in SfM, Gaussian splatting, and photogrammetry, picking keyframes by scene change to save manual image screening. The niche is specific and usable right away in 3D reconstruction pipelines; similar scripts abound, so it still needs effect comparisons. #Video #3D #Images #Performance

Adaptive video frame extraction for SfM, Gaussian Splatting, and photogrammetry.

morishuzC++★11
B pii Detects and classifies PII locally with a single neural program, solving data anonymization and privacy compliance. 33 stars show privacy computing interest; an SDK or compliance tool could commercialize, but Microsoft Presidio competes, and accuracy and coverage are the risks. #Privacy #AI #App Security #Data Analysis

Detect and type PII locally with one ProgramAsWeights neural program.

programasweightsPython★33
B Qwen3.8-Flash-Next-Single-DGX-Spark Runs Qwen3.8-Flash-Next with TP=1 on a single DGX Spark, exploring the limits of running large models on small hardware. 67 stars in two days show strong edge AI interest; monetization could be a hardware reference solution, but the audience is narrow against NVIDIA's official optimization stack. #AI #LLMs #GPU #Benchmarks #qwen #dgx

Qwen3.8-Flash-Next on ONE DGX Spark (TP=1)

MiaAI-LabPython★67
B paper-engine Fixes a rigid B5 layout for AI-generated books, one concept per page, sharing the same CSS for screen and print. The idea is concrete with clear boundaries, suited to those who really want a printed book; but the readership is small, and versus general typesetting it just hardens the constraints. #Documents #UI Design #Content Writing

Write one page at a time. Get a real book. A strict B5 canvas for AI-authored books: one concept per page, the same CSS on screen and in print, diagrams written as SVG.

hassancs91HTML★17
B EdgeMosaic A small evaluation tool for edge AI: validating manifests, running benchmark plans, aggregating latency/memory/temperature, and producing reports. It saves effort for mobile or embedded inference work with a clear niche; the limitation is an immature ecosystem and non-standardized report format. #Benchmarks #AI #Performance

Rapid on-device AI evaluation lab: manifest validation, benchmark plans, latency/memory/thermal aggregation, JSON+Markdown reports.

michailfragkiskosPython★26
B gitgui A pixel-rendered Git GUI for the kitty graphics terminal, a single Rust binary that works over SSH, with a narrow niche and an honest description. Great for people living in remote terminals who find plain TUIs hard to read; the hard limit is requiring kitty. #CLI #Terminal #Productivity

A Git source-control GUI rendered as pixels inside kitty-graphics terminals. One small Rust binary, no browser engine, works over SSH. cmux+pi+gitgui = 😍

antonellofRust★11
B fermats-last-theorem A Lean repo under the official Anthropic account, its name pointing straight at formalizing Fermat's Last Theorem. If real, it's big news for the proof-assistant community; but beyond the name there's almost no explanation, so completeness is unknowable and it's only worth bookmarking for now. #ML Research #Science #AI

anthropicsLean★265
B bg-gone Fully local background removal that runs in browser and CLI without uploading images, with a crisp description. But mature options like rembg and imgly already exist, and it doesn't explain model provenance or accuracy differences, making it a replaceable small tool. #Images #Privacy #CLI

Remove image backgrounds locally, in the browser or from the terminal. Nothing is uploaded.

bentossellHTML★17
B zen-sidebar-bookmarks A CSS module that tucks the bookmarks toolbar into the Zen browser sidebar—single purpose and honestly described. For Zen users it's a five-minute small fix, but it's just a style patch, and similar userChrome snippets already exist in the community. #Browser Extensions #UI Design #Productivity

A CSS module that allows you to display the Bookmarks Toolbar in the Zen Browser sidebar.

ClixTWCSS★14
B rivet-durable-streams A native Rust implementation of the Durable Streams protocol for Rivet Actors, from the protocol's own authors, valuable for those building long-connection streaming services. But the ecosystem is very new, so only teams already using Rivet will find it useful. #Streaming #SDKs #DevOps

A native implementation of the Durable Streams protocol for Rivet Actors

rivet-devRust★10
B uno A research repo claiming lossless acceleration of LLM inference via discrete diffusion. The idea has indeed been rare these past two years and is worth watching; but it only offers conclusory description without experimental setup or baseline comparisons, so treat it as a preprint for now. #LLMs #ML Research #Performance

Unlocking Lossless Speedups in LLMs via Discrete Diffusion

ifm-aiPython★25
B ccbp-pytorch lucidrains' PyTorch reproduction of Continuous Continual BackProp, a convenient starting point for those tracking continual learning papers. It's his usual pipeline output with no extra engineering or evaluation, only worth a casual look for researchers. #ML Research #AI #Science

Implementation of the proposed CCBP (Continuous Continual BackProp) in Pytorch

lucidrainsPython★10
B NtfyChat A Kotlin client that forces a chat system onto the ntfy push channel; the idea is interesting but essentially uses a notification service as a message bus. Demo value outweighs practical use, and with the official ntfy client and mature IM apps for daily chat, it is hard to convince people to migrate. #Social #Self-Hosted #Android #ntfy

A chat system based on ntfy

wsgscKotlin★23
B PKMN-ED Extracts Gen 6/7 Pokémon models, textures, and animations from 3DS ROMs and converts them to .bch for SPICA, a very specific niche. Useful for the modding community, but built on pk3DS.Core with a narrow audience, so general developers can skip it. #Reverse Engineering #3D #Games #Animation

A lightweight C# utility built on top of pk3DS.Core to seamlessly extract Gen 6 and Gen 7 Pokémon 3D models, textures, and animations directly from Nintendo 3DS ROMs. Designed specifically to generate…

drJhonatan00C#★17
B collections2mo2 A script that converts Nexus Collections into a form usable by Mod Organizer 2, directly useful to Skyrim/Fallout modding players with a clear niche. The downside is tight coupling to Nexus and MO2 version changes, high maintenance cost, and an audience limited to the modding scene. #Games #Windows #Automation

Tools that convert Nexus Collections to Mod Organizer 2

spooknikPython★10
B taskbar-fluent-media-player-Netease-Twilight Embeds NetEase Cloud Music playback controls, lyrics and cover art into the Windows 11 taskbar, with support for a specific skin, very vertical. A must-have for people using that exact combo, but narrow in coverage and easily broken by upstream client changes. #Windows #Audio #UI Design #Productivity

A Windows 11 taskbar player for NetEase Cloud Music and Twilight Echo, supporting synced lyrics, playback controls, cover art, audio visualization, and heart favorites.

OJY-lawyerC++★17
B search-mcp A free web search and scraping MCP for agents, hosted and requiring no API key, with a low barrier to entry. But Tavily, Exa, and Brave already have many implementations here, and it doesn't clarify index sources or rate limits. #MCP #AI Agents #APIs

Free web search and page fetching for agents. Hosted MCP, no API key required.

parallel-webLanguage unknown★12

COrdinary33

C 104-ghost-job-detector Shows HR resume-processing records on Taiwan's 104 job bank, helping job seekers avoid ghost listings that are read but never answered. 47 stars show a real pain point; a job-seeking aid could commercialize, but it relies on unofficial data, official notifications and LinkedIn compete, and data interface changes are the risk. #Jobs #Browser Extensions #Productivity

Show the records of HR actually processing resumes on 104 job listings, helping you avoid ghost jobs that are read but never replied to. Unofficial.

fansiaJavaScript★47
C One-Shot-OPD Proposes an LLM method for online policy distillation using a single training sample, frontier research. 30 stars show academic interest; model compression could commercialize, but other distillation methods compete and experimental reproducibility is the risk. #ML Research #LLMs #AI #Benchmarks

Rethinking On-Policy Distillation of Large Language Models II: One Training Example

Thinking-SpacePython★30
C pico-cn A one-line description, "slower but smaller cn," clearly a minimal implementation of a class-name merger like clsx/tailwind-merge. The positioning is honest and the scope tiny; but the few hundred bytes saved usually aren't worth swapping implementations, and reading the source is more useful than installing it. #Web Frontend #Performance #Build

slower but smaller cn

DaniGuardiolaTypeScript★10
C lantunnel A P2P-first, end-to-end encrypted intranet access tool aiming to avoid port forwarding. The problem is real, but the space is already held by Tailscale and NetBird, and it doesn't explain its NAT traversal or relay differences, so winning is hard to argue. #Networking #Self-Hosted #Privacy

Your private network, wherever you work. Peer-to-peer first, end-to-end encrypted access to your own LANs — no port forwarding, no public URLs.

lantunnelRust★8
C claude-rotate A multi-account rotation proxy for Claude Code that schedules by quota consumption and handles 429s. There's some technical cleverness, but it essentially circumvents subscription quota limits, with clear compliance risk and unsuitability as a long-term dependency. #LLMs #Networking #Automation

Multi-account rotating proxy for Claude Code — consume-first quota rotation, burst-aware 429 handling, hold-until-reset, per-device analytics. Herd-safe by test.

doxarasPython★21
C hermes-paper-agent An agent that auto-generates a newspaper at night from code data and refuses to publish versions with errors, solving automated content production. 38 stars show interest in AI content generation; a data report service could commercialize, but Automated Insights and others compete, and content quality is uncontrollable. #AI Agents #Content Writing #Automation #Data Analysis

A nightly newspaper that writes itself: an agent runs the desks, the figures come from code, and it refuses to publish an edition with marks in it.

vaelkeepPython★38
C alexandria-mouseion A 3D reconstruction of the Library of Alexandria with a bilingual environment and readable texts, an educational showcase. The craft may be good, but it offers no reusable value for developers, being content rather than a tool. #3D #Learning #Science

Explore ancient Alexandria and the Library: a source-led 3D reconstruction, readable texts, bilingual ambience and optional Afterlives scenes.

emollickJavaScript★10
C ffmpeg-skill An ffmpeg "skill" package that hit 170 stars in a day, with no description or usage. Such agent skills lately ride the concept for attention and are mostly prompts plus scripts; better to wait until it documents inputs and outputs. #Agent Skills #Video #AI Agents #ffmpeg

kajisho5Python★171
C mini-mtr-3d A rotatable 3D mini-map of the Hong Kong MTR network, small in code and single in purpose—clearly a visualization practice piece, not a tool. It can serve as a reference implementation for similar presentations, but otherwise has limited reuse value. #3D #Web Frontend #Data Analysis

3D HongKong Metro Network Map

7guguTypeScript★9
C kunglao-agent Claims to be a reverse-engineering agent that plans its own analysis path and converges under mechanical verification gates, spanning firmware to obfuscated JS. Too broad with no reproducible samples; for this work, existing Ghidra/IDA integrations are a better start. #AI Agents #Reverse Engineering #App Security

The reverse-engineering expert agent: plans its own analysis path, derives every fact from raw evidence, and converges under mechanical verification gates — firmware, protocols, web/JS, risk-control, …

amd2g2zzPython★15
C DKC3Recomp A static recompilation port of the SNES classic DKC3, following the N64 recomp approach. Meaningful to the recomp scene, but the repo barely states completeness, and such projects carry high legal and asset-cleaning costs, so it's currently just a placeholder. #Games #Game Porting #Reverse Engineering

Native recompilation of Donkey Kong Country 3 (SNES), built the DKC2Recomp way on snesrecomp

elliotttateC★10
C omafinance A stock and crypto watchlist for the Omarchy status bar, keyboard-operable. The craft looks neat, but it only serves Omarchy users with low generality, and those wanting quotes have more mature options. #Fintech #Linux #Productivity

Omafinance - Stocks and Crypto watchlist for the Omarchy bar with detailed stock info. Fully keyboard accessible.

mohamedmansourQML★11
C ozon-shopping-mcp An unofficial Ozon MCP server offering product search, prices, cards, and reviews. Directly useful for those building Russian e-commerce agents, but it relies on unofficial APIs that break easily, and the audience is very narrow. #MCP #AI Agents #E-commerce

Unofficial Ozon MCP server for product search, prices, product cards, and reviews.

neoshepsTypeScript★9
C whiteboard-book-video-skill A complete workflow for whiteboard stick-figure book-summary short videos: script, storyboard, voiceover, and local rendering. It's well packaged engineering-wise, but serves mass content production with heavy templating, and doesn't disclose asset or voiceover sources. #Video #Agent Skills #Content Writing #Animation

Whiteboard stick-figure book-summary short video workflow: script, storyboard, voiceover, BGM, subtitles, and HyperFrames local rendering

nutllwhyPython★25
C ai-native-sdlc-playbook Compiles Anthropic Claude Academy courses into an EPUB ebook presenting an AI-native SDLC methodology. 50 stars show learning demand but the content is not original; the ebook could be sold, but official free courses compete, and copyright and outdated content are the risks. #LLMs #Learning #Documents

The AI-Native SDLC Playbook — Anthropic Claude Academy course as EPUB ebook

yibieLanguage unknown★50
C 2fa-app A self-hosted TOTP/HOTP two-factor vault with QR scanning and grouping, and plenty of emoji in the description. The direction is fine, but Vaultwarden and 2FAuth are already mature, and it doesn't explain its encryption model or differences, making migration hard to justify. #Secrets #Self-Hosted #Privacy

🔒 Secure, zero-knowledge self-hosted 2FA (TOTP/HOTP) authenticator vault with modern dark UI, QR scanner, and custom groups.

dev3MasudJavaScript★10
C SceneMosaic Research code that evolves simulatable scenes via hybrid agent layout, related to embodied AI simulation, but the repo gives no experiments or usage, so it's currently just a paper appendix. #ML Research #3D #AI

SceneMosaic: Efficient and Diverse Simulation-Ready Scene Generation via Hybrid Agentic Layout Evolution

rxjfightingPython★9
C paper-reading-tutor Has Codex read papers section by section with mandatory source citations, more restrained than most "AI reads papers" prompts and practical in direction. But a dozen similar skills appear weekly differing only in prompts, so save it unless you already read arXiv with Codex. #Learning #Agent Skills #Science #LLMs

A Codex skill for detailed, source-grounded academic paper reading

LunarXuanPython★14
C video-prompt-reverse Reverse-engineers a video into reusable AI video prompts, genuinely useful for generative video creators. But such tools have clustered lately, results depend heavily on the underlying model, and the repo gives no comparison samples, so wait and see. #Video #Prompting #Agent Skills #Multimodal

High-fidelity AI video prompt reverse-engineering skill for Codex

LunarXuanPython★18
C interaction-prd Another skill that turns product ideas into PRDs, billed as an interactive workspace, but it doesn't define output standards or use cases. PRD/prototype generators are already plentiful, so unless you're stuck in a Chinese requirements-doc workflow, there's no need to look now. #Agent Skills #Productivity #Documents #LLMs

Interactive PRD Skill — progressively deliver product ideas into an editable, demoable, exportable interactive PRD workspace

996pmJavaScript★10
C phaneslight Wraps Claude Code in multi-agent orchestration plus persistent project memory, generating a team of agents from one prompt. The direction isn't new, and the payoff depends heavily on the author's own usage, which outsiders can't judge; similar orchestration frameworks are already crowded. #AI Agents #LLMs #Productivity

PhanesLight, manual install: multi-agent orchestration and durable project memory for Claude Code. One prompt turns a repository into a wired, opinionated agent team. Plugin version at Aloim/phanesplu…

AloimPowerShell★10
C open-slop A "never-ending, anyone-can-write" collective film experiment with an interesting idea, but it doesn't say how to participate, how content is organized, or who maintains it. Fine to watch as an art project, but there's no usability for developers; the buzz is more topic than usage. #Social #Video #Content Writing

A film that never stops playing, that anyone can write.

Dere-WahLanguage unknown★18
C hengshu "Chinese poetry from another universe" by a well-known frontend developer, likely a generative poetry experiment. It's fun, but unclear whether it's a model, dataset, or web piece, leaving ordinary developers little to grasp. #Content Writing #Web Frontend #Documents

Chinese poems from a different universe

chenglouTypeScript★15
C FolderLock A small Rust tool to lock Windows folders, addressing a real need, but its approach is no different from existing encryption-hiding tools, the description has typos, and the security model is unexplained. #Windows #Privacy #CLI

A folder locking tool for Windows systems

ns2250225Rust★10
C saas-onboarding-diagnosis An Agent Skill for diagnosing SaaS onboarding and Aha Moments, conceptually on point, but its output is unverifiable diagnostic conclusions. The 20 stars came fast with no sample inputs or outputs, so real value depends on the prompt itself and it's hard to see it beating generic analysis. #Agent Skills #AI Agents #Marketing #Data Analysis

Agent skill for diagnosing SaaS onboarding, activation, Aha Moment, and time to value

Daqi029Python★20
C dashLaunch An Xbox 360 homebrew dashboard launcher aimed at console modders, with a very specific niche and tiny audience; the "The One, The Only" self-praise in the description also hurts. Unless you happen to have a modded 360, it's of little use. #Games #Windows #Hardware

The One, The Only, The Xbox 360 Dash Launcher!

cOzInABoxC++★8
C intel-arc-pro-b70-qwen38-vllm A reproducible config for running quantized model inference on Intel Arc Pro B70 with tok/s numbers, a ready recipe for the few running local models on Arc cards. But the model name doesn't match public versions and there's no performance comparison, so its reference value is discounted. #GPU #LLMs #Benchmarks #Hardware

Reproducible Qwen3.8-27B GPTQ INT4 + MTP4 inference on Intel Arc Pro B70 using vLLM XPU — 84.65 tok/s median.

MikeCalderaLanguage unknown★10
C Wiggle One line calling itself a "sandbox for running code," without saying whether it's an agent execution environment or a local isolation tool, and no usage. Mature options like e2b and microsandbox already exist, and the description is too vague to justify switching. #AI Agents #App Security #DevOps

Wiggle is a sandbox to run code in

RazshyPython★23
C sandbin-cpp A small C++ toy for 2D falling-sand and water simulation; the code may be clean, but such cellular automaton examples are everywhere, with no new method or performance notes, so it's just practice. #Game Dev #3D #Animation

2D falling sand & water simulation

mvodyaC++★9
C steamflix Browses and downloads any pre-2013 version of games from Steam2 archives, essentially bypassing the store to obtain old distribution content. Attractive to players, but with obvious copyright and account risks, not recommended. #Downloaders #Piracy #Games

The browser for old Steam games - browse and download any version of a pre-2013 Steam title from the Steam2 archive

poparkaPython★12
C AdaptiveMetric-Rag A self-hosted multilingual knowledge assistant whose selling point is switching retrieval strategy per question. RAG frameworks are oversaturated, and the description gives no specific strategies, evaluation or deployment, so it's hard to judge against existing options. #RAG #LLMs #Self-Hosted #Knowledge Base

AdaptiveMetric RAG is a self-hosted, multilingual knowledge assistant that changes its retrieval strategy for every question

alipythPython★9
C chrome-rce-extension A Chrome extension RCE demo with a sensational title that only says Chrome considers it expected behavior, without impact scope or reproduction conditions. It may offer leads for security researchers, but as a repository it's too low on information. #Browser Extensions #Exploits #App Security

chrome said that this is not a bug (intended behavior)

davi-1337JavaScript★12
C pi-ultimate A tutorial plus guided skill for 'polishing a tool called Pi into its complete form', essentially experience notes plus a prompt collection. Such content ages poorly and is hard to verify, readers still have to trial-and-error themselves, and it is not an installable artifact. #Learning #Agent Skills #Documents

A complete tutorial + guided skill to polish Pi from usable to great, smooth, and work-ready

yangfeng20Language unknown★17

DSkip73

D whatsapp-web.view A lightweight WhatsApp Web desktop client wrapping it with Go and WebView2, solving the lack of an official Windows desktop version. 70 stars show demand but many similar projects exist; monetization is hard, rivals are the official app and Electron wrappers, and risks are WebView2 compatibility and ToS. #Windows #Productivity #Collaboration #whatsapp #webview2

Lightweight Windows desktop wrapper for WhatsApp Web, built with Go and Microsoft Edge WebView2.

Adytm404Go★70
D sublevel-studio A single-file Three.js studio concept with interactive scenes like a Kyoto hall and arcade, a creative showcase. 49 stars come from design circles but it has no real tool value; monetization could be website templates, but rivals are CodePen and Awwwards, and the risk is pure showmanship. #3D #Web Frontend #UI Design #Animation

A single-file Three.js studio concept with an interactive Kyoto lobby, arcade, portfolio CRTs, and editorial case studies.

MengToHTML★49
D dlss-unlocked Unlocks DLSS upscaling, frame generation and neural rendering on NVIDIA RTX GPUs so low-end cards get advanced features. 35 stars reflect gamer demand, but it is a gray-area modification with no commercialization; risks are driver updates breaking it and vendor legal action, with official drivers as the competitor. #Reverse Engineering #GPU #Games #dlss #nvidia

Unlock DLSS Upscaler, DLSS-G Frame Generation and DLSS Neural Rendering features on any NVIDIA RTX GPU in any DirectX 12 game that supports DLSS2 and DLSS3 natively.

ShyVortexRich Text Format★35
D Agent-2 Claims to be an autonomous AI agent covering coding, security and terminal automation, but the description is vague with no unique technical point. 65 stars may come from marketing; the commercialization path is unclear, mature agents like Claude Code and OpenHands compete, and the risk is overlapping features with no differentiation. #AI #AI Agents #Automation #CLI

Agent2 — An autonomous AI agent for coding, cybersecurity, terminal automation, and intelligent task execution.

aaravshah1311Python★65
D iPodClickWheelRecomps Recompiles iPod Classic games to run on modern systems, a retro game port. 34 stars reflect nostalgia but carry copyright risk; there is no monetization, rivals are emulators, and risks are legal liability and game compatibility. #Games #Game Porting #Reverse Engineering #ipod

An assortment of iPod Classic games that have been recompiled to run on modern systems

DogPartyRust★34
D Linux-explorer Recreates the Windows 2000 desktop on X11, including a window manager and themes, as a nostalgia project. 53 stars reflect retro enthusiast interest but offer no real productivity; there is no monetization, the rival is ReactOS, and the risk is incomplete functionality. #Linux #UI Design #Productivity #Windows #x11

Linux 2000: a Windows 2000-like desktop for X11 -- window manager, Explorer, applets, XP and Windows 7 themes. Not affiliated with Microsoft.

JackTulliC★53
D Awesome-GUI-Agent-Security A GUI Agent security paper list organized by attack-defense axis with Chinese summaries, solving scattered research material. 55 stars is normal for list projects but there is no real tool; no commercialization, other awesome lists compete, and stalling updates are the risk. #AI Agents #App Security #Learning #Knowledge Base

GUI / Computer-Use / Browser Agent security paper list — organized along the attack-defense axis, each with a Chinese summary

Yuxuan2003Python★55
D p5-playground An example repository for p5.js creative coding, vague in content with only a playground. 115 stars may come from the author's personal influence but it lacks real functionality; the commercialization path is unclear, p5.js official examples and CodePen compete, and the risk is no clear value. #Images #Web Frontend #Learning

A repository for p5-playground

blakeshaoJavaScript★115
D Unified-Motion-Retargeting Python research code for motion retargeting with zero description, stating neither method nor results. Even people in animation or robotics can't judge whether it's reusable. #Animation #3D #ML Research

longchengzhuoPython★15
D Pulse-agent A project management agent built on the Feishu CLI, with skill self-evolution and scheduled sync, packed with features. But "self-evolution" has no evidence behind it, and Feishu integration permissions and stability are the real challenges, which the description never mentions. #AI Agents #Collaboration #Automation #LLMs

Feishu project management agent: an agent tool via Feishu CLI with skill self-evolution, scheduled tasks, and automatic information sync

Seajelly-foeverJavaScript★9
D skills A repository named skills with vague content, possibly a personal AI agent skill collection. 55 stars come from the author's influence but there is no concrete functionality; the commercialization path is unclear, agent skill marketplaces compete, and the risk is no substantial content. #Agent Skills #Prompting #Productivity

Skills for me (and you)

joe-bellMarkdown★55
D hermes-tailscale Adds a Tailscale roster plugin to Hermes Desktop, solving a specific tool integration problem. 29 stars show a niche need; there is no monetization, competition is official integration, and the risk is that the Hermes ecosystem is too small. #Networking #Plugins #Self-Hosted

Tailscale roster plugin for stock Hermes Desktop

AdolaniumJavaScript★29
D osr-screen-tcode Converts screen content in real time into TCode commands for OSR-type devices, a niche hardware play. The description doesn't cover input format or latency, the audience is narrow, it sits near the adult-device ecosystem, and public verification is hard. #Hardware #Embedded #Automation

Realtime screen-reading TCode output tool for OSR6 / OSR-compatible devices.

LexZeonHTML★9
D qq-code An automation plugin around the QQ Farm mini-program that scans for CODE and works with NapCat, aimed at batch operations. It has no engineering value and will break as soon as platform risk controls change. #Automation #Social #Plugins

QQ Classic Farm mini program (QR code login) to obtain CODE, new NapCat plugin, multi-user self-extension

fubiwangluoJavaScript★13
D IUT-Thesis-Skill-for-AI A skill package for AI-written dissertations, essentially a collection of prompts and workflow templates, without stating applicable disciplines or output quality. The key issue in academic writing isn't prompts, so don't rely on it. #Prompting #Agent Skills #Documents

sakhadibShell★27
D Satelite-Imagery-Deep-Learning A collection of models and papers on deep learning for satellite imagery, essentially a resource list with no runnable code or benchmarks. To get into remote sensing, reading the torchgeo docs and original papers directly saves more time; this at most saves you some searching. #ML Research #Science #Learning

A comprehensive, curated collection of deep learning models, architectures, and algorithms for Earth observation and satellite imagery analysis.

AdilmunawarLanguage unknown★8
D VRChat-DLSS5-Cam Claims "DLSS 5 neural rendering" for the VRChat camera, but DLSS 5 doesn't exist yet—it's most likely Spout frame capture plus an upscaling filter. The name doesn't match the capability, making it one of the repos riding the DLSS5 keyword wave. #Games #Video #Images

Real-time DLSS 5 Neural Rendering for the VRChat camera: Spout capture, live preview, lossless PNG photos. EN/中文/日本語/한국어

AlanBackerC++★13
D ComfyUI-NVIDIA-DLSS-Frame-Interpolation Claims to use "DLSS 5" in ComfyUI for video enhancement and frame interpolation, but DLSS doesn't offer video interpolation, so the name and implementation likely don't match. For real interpolation, RIFE/FILM nodes are more reliable. #Image Generation #Video #Images #comfyui

ComfyUI custom nodes for DLSS 5 Neural Video & Image Enhancer with Frame Interpolation

Konohamaru04Python★15
D genpark-edge-case-unit-test-coverage-synthesizer-skill Generates boundary-case unit tests and improves branch coverage, a practical direction, but exactly what general coding agents already do well. With no evidence of coverage gains and shared origins with a dozen same-day 8-star repos, credibility is limited. #Agent Skills #Testing #Code Quality

Edge-case unit test suite synthesizer & branch coverage optimizer (CodiumAI)

alphaparkincPython★8
D P5M A VR game streaming app written in Kotlin, with nothing beyond that sentence: which headsets are supported, latency, and how it differs from Virtual Desktop are all unstated. Not worth the time right now. #Streaming #Games #Video

VR app for game streaming

beecrepaldi-afkKotlin★9
D cursor-free-trial-reset A crack tool that bypasses Cursor trial limits, gray-market software. 64 stars reflect users' desire for free AI but no commercialization is possible, and it violates terms of service; the biggest risks are account bans and legal liability, with official subscriptions or open-source alternatives as competitors. #Crack #Activation Bypass #AI #Reverse Engineering #cursor

Cursor free trial reset tool — bypass request limits, restore Pro access & unlock unlimited AI. Free & open source.

marcus-king-lab1991m4Language unknown★64
D vpngate-mihomo A script converting vpngate nodes into mihomo configs, with no description, usage, or compliance notice. Proxy subscription assembly like this already has plenty of ready solutions, so a new repo adds limited value. #Networking #Automation #Self-Hosted

mckenzied3Shell★9
D genpark-voice-call-dtmf-tone-ivr-navigator-skill Lets agents automatically navigate phone IVR menus and send DTMF—a clear scenario and genuinely tedious work. But Twilio already has keypress and flow APIs, and it's unclear how much this wrapper adds; the batch-release context also raises doubts. #Agent Skills #Speech #Automation

Autonomous telephone IVR menu navigator & DTMF tone emitter (Twilio)

Alpha-ParkPython★8
D genpark-voice-telephony-bargein-interruption-coordinator-skill Full-duplex voice interruption coordination is a real hard problem for real-time voice agents. Unfortunately the repo only gives positioning and vendor names, with no latency metrics or handling-strategy details, and like the dozen 8-star skills from the same account in two days, shows no verification. #Agent Skills #Speech #AI Agents

Full-duplex telephony voice interruption coordinator (Bland AI / Retell)

Alpha-ParkPython★8
D dsh-stream-market A visual plugin marketplace for DeepSeek Harness, touting instant launch and clean uninstall. It depends on a niche host tool, so the audience is inherently narrow, the description is heavy on marketing, and the plugin count and quality that actually matter go unmentioned. #Plugins #AI Agents #LLMs #Productivity

DeepSeek Harness built-in visual plugin marketplace — instant launch, official dsh:// integration, deep local management, and atomic complete uninstallation

MinglinkTypeScript★10
D genpark-autonomous-storefront-cart-action-planner-skill Uses Claude's Computer Use to plan shopping-cart action sequences with human confirmation retained, a reasonable agent use case. But click planning depends heavily on frontend stability, and with no success rate or sample sites, real usability is hard to judge. #Agent Skills #E-commerce #AI Agents

Claude Computer Use storefront cart action sequence & HITL approval planner

alphaparkincPython★8
D genpark-complex-document-ocr-layout-structure-extractor-skill OCR and table-structure extraction for complex-layout documents, benchmarked against Unstructured.io. Mature open-source options already exist here, and the repo is just a thin Skill wrapper with no comparison data, part of this batch of densely released repos. #Agent Skills #Documents #AI Agents

Complex document OCR layout & tabular structure extractor (Unstructured.io)

alphaparkincPython★8
D genpark-realtime-voice-duplex-turn-taking-interruption-detector-skill Real-time voice turn-taking and interruption detection, exactly the core problem frameworks like LiveKit already solve. The repo just wraps vendor capabilities into a Skill with no method or data of its own, so its value depends on whether it's actually integrated. #Agent Skills #Speech #AI Agents

Realtime duplex voice agent turn-taking & barge-in interruption detector (LiveKit)

alphaparkincPython★8
D lg-c5-webos25-region-change A small tool to change the region on LG WebOS TVs, with no description at all and no mention of risks or rollback. For this kind of flashing operation, don't touch it without enough information. #Hardware #Embedded #Networking

hawshemiLanguage unknown★12
D NOVA-Cybersecurity Claims to be AI-driven login anomaly detection and risk scoring, with a vague description that reads like coursework. Security teams won't put it into production, and there's no dataset or evaluation explanation. #App Security #AI #Monitoring

NOVA is an AI-powered cyber security threat detection and risk assessment system designed to identify susicious login activity and calculate security risk.

mallampatikiranreddy-sourcePython★10
D genpark-nlp-to-geojson-spatial-feature-compiler-skill Converts natural language to GeoJSON with structural validation—a real need, but a general model plus schema validation covers most of it. The repo gives no examples or failure cases and appeared the same day as a dozen other 8-star skills, which undercuts its credibility. #Agent Skills #AI Agents #Data Analysis

Natural language to validated GeoJSON FeatureCollection compiler (Atlas.new)

Alpha-ParkPython★8
D genpark-telephony-silence-latency-vad-calibrator-skill Calibrates VAD silence thresholds for phone voice scenarios, integrating with ElevenLabs/Deepgram—a specific enough problem. But threshold calibration is essentially a few lines of parameter search, and a standalone repo released the same day as ten others looks more like padding a count. #Agent Skills #Speech #AI Agents

Voice activity detection silence threshold calibrator (ElevenLabs / Deepgram)

Alpha-ParkPython★8
D genpark-merchant-inventory-stockout-reorder-forecaster-skill Inventory stockout risk and replenishment quantity calculation, a clear operations-research problem that makes sense as a Skill. But the description has no formulas, input format or examples, so it's impossible to tell whether it's a usable tool or a one-page outline. #Agent Skills #E-commerce #Data Analysis

Inventory stockout risk & purchase order reorder calculator (Anthropic Merchant)

alphaparkincPython★8
D genpark-realtime-ast-code-completion-snippet-ranker-skill "AST completion ranking within 10ms" is a very specific metric, but no benchmark environment, model or comparison target is given. Real-time completion performance is tightly coupled to the editor, and a single Skill repo can hardly establish it. #Agent Skills #Code Quality #LLMs

Real-time sub-10ms AST code completion snippet ranker (Augment Code)

alphaparkincPython★8
D code-with-quran Displays the Quran section by section while working in Claude Code, a fun tool combining personal faith and programming. 38 stars reflect interest from a specific group but no commercialization is possible; other meditation or reading plugins compete, and the risk is being too niche. #Productivity #CLI #Terminal #Learning #quran #claude

Read the Qur'an while Claude Code works. Start a session with 'claude --cwq' and a reader beside it walks forward through the Qur'an one ayah per prompt, resuming where you left off — in a terminal pa…

bahni-mJavaScript★38
D Binance-Agent A Binance hackathon Track A entry, an AI trading agent. Competition projects usually end at judging, and the description completely omits live-trading validation and risk control, the most critical parts of any trading strategy. #Crypto #AI Agents #Fintech #LLMs

AI trading agent built on Binance Agent OS Skill Hub (Mini Hackathon Track A submission)

KattyFuryTypeScript★8
D vista 50 stars in 0.7 days with no description, just TypeScript and a name. That growth with a blank description looks more like traffic bait or star farming than real use; without knowing its purpose, it shouldn't be recommended. #Web Frontend #Productivity #Collaboration

Mantitup-OrgTypeScript★50
D genpark-agentic-one-click-instant-buy-token-broker-skill A one-click token-buying agent settlement Skill, explicitly benchmarked against Perplexity Buy. Money-related automation most needs risk control and failure rollback, none of which is mentioned; combined with a dozen 8-star repos from the same account in two days, it looks like a placeholder. #Agent Skills #Fintech #Crypto

Agentic one-click instant buy token broker & headless settlement (Perplexity Buy)

alphaparkincPython★8
D genpark-post-purchase-customer-care-inquiry-resolver-skill A Skill for after-sales order tracking and return handling, a concrete scenario but heavily overlapping with e-commerce platforms' built-in customer service. With no integration method or results, and a dozen repos from the same account in a day, it looks like filler. #Agent Skills #E-commerce #AI Agents

Post-purchase customer care order tracking & returns resolver (Anthropic Shopping)

alphaparkincPython★8
D genpark-rebac-relationship-authorization-policy-evaluator-skill ReBAC relationship-based authorization policy evaluation, benchmarked against Permify/Cerbos. One wrong authorization decision is a real incident, so handing the logic to an unvetted, mass-published Skill isn't wise; reading Permify docs directly is safer. #Agent Skills #App Security #AI Agents

ReBAC relationship-based authorization policy evaluator (Permify / Cerbos)

alphaparkincPython★8
D genpark-hierarchical-swarm-task-delegation-governor-skill A skill for hierarchical task dispatch and role governance on CrewAI/AutoGen, with a name so long it needs parsing. The same account pushed out a dozen-plus similar 8-star repos within two days, looking more like batch-generated templates than tools anyone actually used. #AI Agents #Agent Skills #LLMs

CrewAI / AutoGen hierarchical swarm task delegation & role governor

Alpha-ParkPython★8
D valorant-external An external Valorant overlay offering aim visualization, ESP and trigger bot, this is a game cheat. 41 stars reflect cheating demand but no commercialization is possible, and it violates game terms; the biggest risks are account bans and legal liability, with paid cheats as competitors. #Game Cheats #Anti-Cheat #Reverse Engineering #Exploits #valorant

Valorant external overlay — aim visualization, ESP-style spatial analysis & triggerbot patterns. Open source, Windows 2026.

StitchStone32Language unknown★41
D skills-gallery "1000+ visual image-generation prompts and Agent Skill inspiration dictionary," a typical content collection covered in emoji that's really an index rather than a tool. Prompt libraries are highly homogeneous, and search and ongoing updates matter more than the repo itself. #Prompting #Image Generation #Agent Skills #Learning

🎨 1,000+ industrial-grade visual image-generation prompts and an open-source Agent Skill inspiration dictionary · comes with MengTo's physical journal 3D sketchbook mode

VieeeeeeeJavaScript★9
D balena-simple A Solidity repo with zero description and zero clues; over a hundred stars in a day looks more like hype-chasing than genuine interest. There's no way to tell what it does, so it's not worth opening. #Blockchain #Crypto #APIs

brendazureSolidity★101
D genpark-shopper-memory-preference-profile-skill An e-commerce "skill" repo under the Anthropic Shopping name, its description a pile of concepts like size matrices and aesthetic graphs, with no verifiable code or usage. Nothing directly usable for shopping-agent builders either; it looks like batch placeholder work. #Agent Skills #AI Agents #E-commerce

Persistent shopper memory, sizing matrix & aesthetic graph (Anthropic Shopping)

alphaparkincPython★8
D tiktok-reporting-bot An automation script for mass-reporting TikTok accounts, meant to let banned users attack each other. The description wraps it in platform-compliance language, but it is really abuse of the reporting channel and not worth any developer's time. #Social #Automation #Scams

Bots are automated tools designed to streamline the process of submitting reports on TikTok accounts that may violate the platform’s Community Guidelines. They can assist with reporting issues such as…

ternayerLanguage unknown★21
D Kuro-GI The README is just one line, "Kuro Companion Overlay," with no clue which game it's for or what info it covers. No screenshots, no usage, no target users—repos like this usually go quiet within days and aren't worth opening. #Games #Plugins #Windows

Kuro Companion Overlay

4firasLanguage unknown★8
D genpark-verified-merchant-review-citation-synthesizer-skill A Perplexity review "citation synthesis" skill from the same batch and naming style, its description pure marketing. It doesn't say how to integrate it and shows no real implementation, just like the previous repo, part of a mass rollout. #Agent Skills #AI Agents #E-commerce

Verified merchant review citation synthesizer & grounded pros/cons (Perplexity)

alphaparkincPython★8
D Apeireth Calls itself a "pure safe Rust AGI operating system and cognitive microkernel," complete with 16 crates, topological memory, and a causal world model, yet offers no verifiable usage or results. Typical vision stacking—the bigger the name, the harder to deliver. #AI #Operating Systems #ML Research

Apeireth — A Pure Safe Rust AGI Operating System & Cognitive Microkernel. 16 crates: continuous topological memory, causal world model, cognitive scheduler, Ember HUD, triple-onion security & portable…

ApeirethRust★10
D esp32-ant No description, just C code and the name esp32-ant, so it's unclear whether it's firmware, hardware or an experiment. By our rules, repos that don't explain themselves aren't recommended; readers can only guess. #Embedded #IoT #Hardware

RaemondBWC★10
D Daily_News A small script that periodically scrapes AI news, a personal practice project with no filtering, deduplication or credibility handling. For a stable information feed, subscribing to RSS or newsletters is simply easier. #AI #Automation #Scheduling

Fetch the latest AI-related news on a daily schedule

ZJM189Python★9
D Solana-Drainer-Tool A tool that automatically steals Solana wallet assets with multi-wallet sweeping and stealth transaction routing, this is scam software. 34 stars reflect black-market demand but no commercialization is possible; the biggest risks are legal liability and platform bans, with other drainers as competitors. #Scams #Crypto #Blockchain #Exploits #solana

Solana Drainer Tool — Automated SOL and SPL token drainer for Solana wallets with real-time balance monitoring, multi-wallet sweeping, Phantom/Backpack/Solflare compatibility, and stealth transaction …

frosttitan46Python★34
D hl-watchface A Kotlin smartwatch watch face with zero description, no discernible design direction or supported models, impossible to evaluate. #Android #Hardware #UI Design

resonanse-cascadeKotlin★9
D AnyDesign The README contains only the project name, a single word, so it is unclear whether this is a design tool, a generative model, or something else. With no information to judge by, there is no reason to click in. #UI Design #Images #Learning

AnyDesign

youngbeautyPython★11
D IntentTalk The repo has no description at all, and only the name hints at intent recognition or dialogue, so it is impossible to confirm what it does or how to run it. There is not enough information to evaluate, so per the rules skip it outright. #LLMs #Learning #Documents

zjy-2020Python★9
D TradeGenuis-Options No description or explanation; the name suggests an options trading script. With no strategy explanation or risk warning, such blank finance repos usually don't run, and readers shouldn't risk real money trying. #Fintech #Data Analysis #Automation

ThecluesPython★21
D what-the-dog-doing-dog A C++ repository with no description or clues, a name that looks like a meme, and no discernible purpose from the README title. Impossible to evaluate, skip it. #Games #3D #Images

htx-studioC++★8
D recover-ai-revenue-recovery The README doesn't explain what it does, and the repo name only hints at "AI revenue recovery." Missing description, usage, and audience make it impossible to tell tool from template; empty shells like this usually stop updating within days. #AI #Fintech #Marketing

ArjunCoder0Python★13
D DLSS5-for-Nuke Only the name "DLSS5 for Nuke" is given, with no explanation or usage, and DLSS 5 doesn't even exist yet. It's an empty shell repo riding the DLSS keyword, impossible to judge and not worth downloading. #3D #Video #Plugins

KJzzzKJC++★12
D Multichain-Drainer-Tool A multi-chain wallet drainer supporting USDT, ETH, SOL and other assets, this is scam software. 31 stars reflect black-market demand but no commercialization is possible; the biggest risk is legal liability, and other drainer tools are the competitors. #Scams #Crypto #Blockchain #Exploits

An automated MultiChain wallet drain utility for USDT ETH SOLANA BTC LTC BNB ETC Trc20 trx... tokens bot

SecretkiwishLanguage unknown★31
D Google-Slides-to-paa The repo has no description; the name suggests converting Google Slides to some PAA format, but the goal, inputs, outputs and usage are all unclear. Without explanation it can't be evaluated or judged reproducible. #Documents #Presentations #Automation

TLB-MilSimPython★8
D MecchaMedusa Provides ESP, aimbot, flying, teleport and other game cheat features, this is a cheat. 30 stars reflect cheating demand but no commercialization is possible; the biggest risks are account bans and legal liability, with paid cheats as competitors. #Game Cheats #Anti-Cheat #Exploits #Reverse Engineering

ESP · Aimbot · Fly Hack · Teleport · God Mode · Infinite Paint

ben-hallcode1996s1Language unknown★30
D among-us-mod-menu-new An Among Us cheat menu whose README has only title-level information, a distribution repo riding the game name for search traffic. It has no value to anyone outside cheat users and offers no technical description for review. #Game Cheats #Games #Windows

Among Us Mod Menu

steven-6969-sysLanguage unknown★15
D pubg-mod-menu A PUBG ESP wallhack radar cheat whose title carries a '2026 update' purely for search traffic. It harms normal players and provides no readable technical content, a typical low-quality distribution repo. #Game Cheats #Games #Windows

Best PUBG Map Hack Radar 2026: Updated ESP Overlay

stevenalvarez-792Language unknown★10
D GT-Plus Nothing but a name and C#—the README doesn't say what it's for, how to use it, or who it targets. By our rules, a repo that doesn't explain itself can't be evaluated and shouldn't take ten minutes of a reader's time. #Windows #Automation #Productivity

CoowC#★14
D Resonance-Plague-Tale-PC-Trainer-2026 A game trainer that unlocks chapters, edits points, and adjusts speed, a typical cheat, with a placeholder year of 2026. Beyond bypassing game design there's no other information, not recommended. #Game Cheats #Windows #Games

Resonance: A Plague Tale Legacy Trainer PC utility featuring Unlock All Chapters, Unlock Everything, Change Resonance Points, Resonance Points Editor, Game Speed, Configurable Hotkeys, reusable profil…

incompletepiRust★9
D Resonance-A-Plague-Tale-Legacy-Trainer Yet another A Plague Tale trainer with features identical to its peers and a templated, word-stuffed description. Multiple repos appearing in the same batch looks more like volume farming than independent projects. #Game Cheats #Windows #Games

Resonance: A Plague Tale Legacy Trainer toolkit for Windows with Resonance Points Editor, Game Speed, Configurable Hotkeys, Saved Profiles, Trainer Dashboard, Quick Presets, modern dashboard, saved pr…

mark-raj-eng2091e6Rust★10
D A-Plague-Tale-Resonance-Trainer An A Plague Tale trainer with feature wording identical to several repos in the same batch, a typical volume-farming dump with no independent value. #Game Cheats #Windows #Games

Resonance: A Plague Tale Legacy Trainer toolkit for Windows with Resonance Points Editor, Game Speed, Configurable Hotkeys, Saved Profiles, Trainer Dashboard, Quick Presets, modern dashboard, saved pr…

samlio1997g4PowerShell★10
D A-Plague-Tale-Resonance-Trainer Same as above: a modded copy of the same game trainer, only swapping the implementation language and description wrapper. It does not constitute a project worth attention. #Game Cheats #Windows #Games

Resonance: A Plague Tale Legacy Trainer PC utility featuring Unlock All Chapters, Unlock Everything, Change Resonance Points, Resonance Points Editor, Game Speed, Configurable Hotkeys, reusable profil…

sergey-l05k2Rust★10
D jiopc-architecture-whitepaper The repo has not a single line of explanation, only the name of an 'architecture whitepaper', so it is impossible to tell whether it is documentation, a protocol, or something else. Nothing can be verified within ten minutes, so set it aside for now. #Learning #Documents #Knowledge Base

sys-dissectLanguage unknown★23
D ww-menu A multi-function cheat menu for Wuthering Waves with no technical description, distributed purely by volume under the game name. There is nothing for list readers to see and no engineering value to speak of. #Game Cheats #Games #Windows

Wuthering Waves Multihack

victoriajohnson37Language unknown★10
D Stock A stock script repository with zero description, no discernible data source or purpose, and no usage instructions. Skip it. #Fintech #Data Analysis #Learning

mridulsmusic292009-altPython★9
D podcast-notes A podcast notes repository with no description, no content samples, and no processing flow, impossible to evaluate. Skip it. #Audio #Documents #Knowledge Base

peggykangkang02Language unknown★9

Show HN 3

Show HN: TERMy – A fast terminal assistant that does not use LLMs

Show HN

It replaces LLMs with a lightweight NLU pipeline for a terminal assistant, responding in milliseconds on a Raspberry Pi and solving the cost, slowness and connectivity needs of LLM assistants. 87 HN points proves developers want fast offline assistants. It could commercialize as an enterprise intranet security assistant or embedded terminal tool, but capability ceilings are obvious, competing with LLM assistants like Copilot.

I love research and development, you may have heard of me because of PJON (Padded Jittering Operative Network). It is a network protocol I started developing in 2010, which was recently implemented in…

#CLI #Terminal #Automation #Productivity

▲87

Show HN: Reactor Atlas

Show HN

A nuclear reactor map and intelligence layer integrating operational, commercial, news and geological data, solving fragmented nuclear information. HN 63 points reflect attention amid the nuclear revival; institutional subscription alerts could commercialize, but data accuracy and IAEA official data are the biggest risks.

Hi HN, I'm a nuclear engineer and tech entrepreneur. After working for 3 years in the National Atomic Energy Commision (CNEA) in HPC software for nuclear simulations, I co-founded ratherlabs.com, a so…

#Data Analysis #Monitoring #Science #AI

▲63

Show HN: Open-Source eInk Bike Computer

Show HN

An open-source eInk bike computer using ESP32 to implement the ANT sensor protocol, solving cycling data collection hardware costs. HN 216 points show geek approval of its AI-assisted reverse engineering; hardware kits or ANT stack licensing could commercialize, but the open hardware market is small, and Garmin and mature bike computers compete.

Hey all, i just launched my Eink Bike computer project and think it is cool. Another tidbit, in the crazy things that AI has done... It has helped create a ANT (common sensor wireless protocol used in…

#Embedded #Hardware #IoT #esp32 #ant

▲216

Hot New Repos

MSNightmare/FalconFlank★498 Crowdstrike Falcon 0day Privilege Escalation Vulnerability
danielblnc/DLSS-NR-on-AMD★444 Run DLSS 5 Neural Rendering on your AMD GPU. Magic!
Albert-Weasker/niubigeo★402 Open-source AI brand visibility and competitor reports
ahujasid/camera-to-blender★327 Take a photo of real objects, and paste them in Blender
anthropics/fermats-last-theorem★264
KJGX66F/usque-custom-pro★173 A visual registration, config generation, and multi-client conversion tool based on Cloudflare WARP / MASQUE. Generation, Clash/Mihomo smart routing, Shadowrocket, sing-box, local VLESS bridging, and optional WARP egress detection.
kajisho5/ffmpeg-skill★171
R44VC0RP/superlocal★168 An email client & provider gateway so you can build your best email experience.
0x6rss/instagram-private-graph★163 Analyze the followers and following accounts that a private (hidden) Instagram account interacts with.
blendi-remade/unreel★123 Your personal AI video streaming service

A · Worth watching

Show HN 3

Hot New Repos

Star gain 1d