@aradotso

vinext ports next.js routing, ssr, and rsc to vite with ~94% api coverage, supporting both routers and native cloudflare workers deployment alongside optional nitro for other platforms.

flipoff-split-flap-display covers building, customizing, and embedding a vanilla js airport-style flip board emulator. supports real-time data feeds, custom messaging, keyboard control, and tv/kiosk deployment with no framework dependencies.

local socks5 proxy that tunnels telegram desktop traffic through websocket connections to telegram data centers, bypassing network-level blocking without requiring external infrastructure.

omnivoice-tts wraps a multilingual zero-shot diffusion tts model supporting 600+ languages with voice cloning from audio, parametric voice design via attributes, and auto-generated voices. includes cli tools for batch inference and web demo.

dingtalk-workspace-cli unifies DingTalk contacts, calendar, todos, attendance, approvals, and chat into a single CLI tool with structured JSON output and built-in agent skill discovery for LLM workflows.

bluehammer-vulnerability-poc covers building, running, and debugging a c-based proof-of-concept exploit in isolated lab environments. includes pgp verification, payload crafting, binary analysis patterns, and troubleshooting for common c vulnerability poc issues.

pyre-code-ml-practice is a self-hosted browser-based coding platform with 68 ml problems spanning transformers, diffusion, alignment, and inference. instant feedback grading via local fastapi backend, no gpu required. covers internals from relu to flow matching with structured learning paths.

legalize-es-spanish-legislation provides queryable access to 8,600+ spanish laws as markdown files with full git history back to 1960, enabling text search, temporal diffs, and reform tracking via standard git and shell commands.

x-mentor-skill-nuwa bundles writing methodology from 6 top X creators into a content OS covering tweet composition, hook engineering, growth mechanics, and account diagnostics with decision heuristics and anti-pattern reference.

gemma-tuner-multimodal enables fine-tuning of gemma 3n and 4 models across text, images, and audio on apple silicon without gpu hardware, using pytorch and metal performance shaders with optional cloud dataset streaming.

my-brain-is-full-crew embeds 10 specialized ai agents into obsidian for knowledge management, wellness, and productivity. agents communicate via shared message board and respond to natural language commands across vault organization, meeting transcription, email integration, and optional health coaching.

deepseek-ocr extracts text from images and pdfs via vision-language inference, supporting multiple prompt modes (document-to-markdown, free ocr, figure parsing, grounding) and two backends (vllm for throughput, huggingface transformers for flexibility).

codesight-ai-context generates compact codebase maps and wiki knowledge bases that feed ai coding assistants in a single context window, supporting 14 languages, 30+ frameworks, and 13 orm parsers with optional mcp server integration.

parlor-on-device-ai is a real-time multimodal voice and vision assistant running entirely locally via fastapi websocket, combining gemma 4 e2b for inference with kokoro tts for output, supporting barge-in and sentence-level streaming without cloud calls.

aracli-deploy-management covers infrastructure deployment patterns (vms, containers, bare metal, serverless, hybrid) and three management approaches (cli, api, mcp) with trade-off analysis but lacks actionable intent and decision procedure.

weft-ai-language teaches building durable ai workflows in rust where llms, humans, apis, and databases compose as typed graph nodes with compile-time validation and automatic visual generation.

binance-futures-signal-bot automates trading across binance, bybit, and okx futures with telegram/tradingview signal ingestion, configurable ema/rsi/macd/bollinger strategies, and real-time p&l dashboards. multi-exchange support and testnet validation included.

keychron-keyboards-hardware-design provides production cad files (step, dxf, pdf) for 50+ keyboard and mouse models with python inventory, conversion, and inspection scripts. covers sparse checkout, batch processing, and 3d printing tolerances.

picoclaw-ai-assistant is a minimal go-based ai chat tool for constrained devices, supporting multiple llm providers and optional web search, deployable across arm, risc-v, mips, and x86 platforms.

openviking-context-database is a filesystem-based context store for ai agents, organizing memories, resources, and skills across tiered load levels (L0/L1/L2) with semantic search, session compression, and observable retrieval trajectories.

sheets-terminal-spreadsheet is a vim-keybinding csv editor for the terminal with formula support, visual selection, and command mode. covers interactive and non-interactive workflows via go or bash integration.

quien-whois-lookup wraps a go-based cli/tui tool for domain, ip, dns, mail, ssl/tls, and tech stack inspection. supports interactive tabbed views and json output for scripting. includes rdap-first lookups with bgp fallback and peeringdb enrichment.

openduck-distributed-duckdb provides distributed query execution across local and remote duckdb workers via transparent attach syntax, with append-only differential storage and hybrid plan splitting via grpc.

ai-scanner-garak wraps nvidia garak in a rails web app for testing llm safety across 179 probes and 35 vulnerability families, with scheduling, pdf reports, multi-tenancy, and siem forwarding.

littlesnitch-linux provides kernel-space network interception via ebpf, with user-space loader and blocklist configuration. architecture spans kernel programs, shared types, and async event consumption from ring buffers.

codex-session-patcher detects and rewrites ai refusal responses across codex cli, claude code, and opencode, optionally injecting ctf-flavored prompts to reduce future refusals via cli, python api, or web dashboard.

web-access-claude-skill extends claude code with browser automation via chrome devtools protocol, combining websearch, webfetch, and cdp proxy to handle login-gated pages, dynamic content, and parallel multi-target research with accumulated site-specific patterns.

agent-flow-visualization renders claude code agent execution as interactive node graphs in vs code, enabling inspection of tool calls, branching, and timing via live hooks or jsonl log replay. supports multi-session tracking and subagent coordination.

lightpanda-browser is a zig-based headless browser for web automation and scraping, offering cdp compatibility and lower resource consumption than chrome, though still in beta with partial web api coverage.

caveman-token-optimizer compresses ai agent responses to 65-75% fewer tokens through pruned prose while preserving code blocks and technical terms exactly. four intensity levels toggle between professional brevity and telegraphic compression.

phantom-ai-coworker is a claude agent sdk-based daemon with persistent memory, dynamic mcp tool registration, and multi-channel communication (slack, email, telegram, webhooks). it can provision infrastructure and self-evolve via separate validation model, but lacks concrete trigger phrases and failure mode documentation.

autoteam-chatgpt-rotation automates chatgpt team account lifecycle management with quota monitoring, account rotation, and codex oauth sync to cliproxyapi. operationally complete but lacks clear intent framing and trigger specificity.

datatalks-data-engineering-zoomcamp is a 9-week free course with code examples for docker, terraform, kestra, bigquery, dbt, and spark. covers production pipeline patterns but lacks prescriptive sequencing and error handling guidance.

cli-to-js-api-wrapper converts command-line tool help text into typed javascript apis via proxy objects, enabling structured invocation of subcommands and flags without shell string concatenation.

jkvideo-bilibili-react-native is a react native client for bilibili with dash playback, danmaku, and live streaming, but lacks actionable procedures for extending or building custom features.

a headless browser automation cli built in rust that wraps chrome devtools protocol for ai agent workflows, supporting navigation, interaction, screenshots, and network interception without node.js.

chrome-cdp-live-browser connects ai agents to running chrome tabs via the chrome devtools protocol, preserving login state and cookies without spinning up isolated browser instances.

kordoc parses korean government documents (hwp, hwpx, pdf) into markdown and structured data, with table extraction and form field recognition. covers cli and programmatic api but lacks operational depth.

724-office-ai-agent is a python-based 24/7 agent with 26 built-in tools, multi-layer memory, and plugin support. lacks actionable procedures and concrete trigger examples for practical deployment.

dflash-mlx-speculative-decoding accelerates inference on apple silicon by using a small draft model to generate token candidates in parallel, verified by a target model in one forward pass, yielding 1.7-4x speedups with lossless output guarantees.

async python bot that identifies polymarket yes/no events, filters non-sports standalone markets, and automatically buys no positions when prices fall below a configured threshold. paper trading by default, live trading requires explicit environment configuration.

designlang-design-extract crawls websites with a headless browser to extract design tokens, colors, typography, and spacing, outputting results as markdown, html, tailwind config, react theme, figma variables, and css custom properties.

anubis-android-app-manager uses shizuku to disable app groups based on vpn connection state, supporting three policies (local, vpn-only, launch-with-vpn) to prevent code execution and network access when apps should be frozen.

codex-autoresearch-loop implements an autonomous modify-verify-retain cycle that iterates toward measurable code goals, reverting failures and stacking wins in git until interrupted or capped.

openclaw-control-center provides a unified dashboard for monitoring openclaw agent activity, token consumption, task execution, and cross-session collaboration with security-first permission defaults.

antigravity-manager proxies google gemini and anthropic claude web session tokens as openai-compatible api endpoints with multi-account rotation and quota tracking. installation and configuration are well-documented, but skill lacks clear intent definition and decision triggers for agent use.

translate-book-parallel chunks large documents into isolated translation tasks executed by parallel subagents in claude code, avoiding context window exhaustion and semantic drift across chapters.

gstack provides eight specialized ai workflows for claude code (ceo review, engineering planning, code review, qa testing, browser automation) summoned via slash commands, but lacks concrete procedures and failure mode documentation.

generates questionnaire-style app onboarding flows by analyzing existing codebase and applying conversion-optimized patterns, outputting screen designs and framework-specific code.

llm-public-opinion-analytics aggregates trending topics from 15 chinese platforms via web scraping, applies llm-powered sentiment analysis and clustering, and distributes alerts across wechat, telegram, and email. covers installation, configuration, and api usage but lacks clarity on decision logic and failure handling.

polymarket-arbitrage-bot automates a dump-and-hedge strategy on 15-minute prediction markets by detecting price drops, buying dipped outcomes, and hedging when combined cost meets profit thresholds.

code-review-graph indexes codebases into a sqlite graph via tree-sitter and exposes them to claude over mcp, reducing token consumption during reviews. lacks concrete procedures and edge case handling.

llmfit matches locally available hardware against hundreds of llm model profiles, recommending which models will execute efficiently given detected cpu, ram, and gpu resources. supports multiple runtime backends and quantization strategies.

inkos-multi-agent-novel-writing orchestrates autonomous novel generation via a writer-validator-auditor-reviser pipeline with human review gates, but lacks concrete procedural depth and operational guardrails.

edict-multi-agent-orchestration models a tang dynasty governance hierarchy as an ai orchestration pipeline with 12 agents, mandatory review gates, kanban dashboard, and audit trails. lacks concrete implementation steps.

memory-lancedb-pro-openclaw wraps lancedb vector storage with hybrid retrieval and cross-encoder reranking for openclaw agents. incomplete documentation leaves implementation and failure handling unclear.

type4me-macos-voice-input captures audio via hotkey, transcribes with local or cloud asr, optionally refines text through llm, and injects into any macos app with full local credential storage.

career-ops-job-search wraps claude code into a job search coordinator that scores offers, generates ats pdfs, and scans 45+ company job portals. lacks explicit workflow triggers and failure recovery patterns.

weclaw-wechat-ai-bridge routes wechat messages to multiple ai backends via a go subprocess, supporting json-rpc, cli spawn, and http modes with qr login and media conversion.

claw-code-harness provides cli and python tooling for inspecting claude code agent manifests, enumerating subsystems, and auditing tool inventories across a python-first implementation with planned rust support.

skillclaw-skill-evolution outlines a multi-agent skill-sharing framework using local proxy interception and cloud-backed skill distillation, but lacks operational depth on the actual evolution mechanism, triggering rules, and failure recovery.

3dsvg-interactive-react converts SVG files into interactive 3D React components via the 3dsvg library and visual editor, leveraging Three.js for rendering.

talk-normal-llm-prompt is a system prompt that reduces ai filler and corporate padding in llm responses while retaining factual content, claiming ~73% character reduction without information loss.

guides writers toward clarity and consistency in chinese technical documentation by enforcing typography rules, avoiding buzzwords, and providing automated linting to catch style violations.

holyclaude-ai-workstation packages claude code cli, web ui, headless browser, and 5 ai clis into a single docker image, promising rapid deployment of a full coding environment without manual tooling.

geoflow-content-automation chains ai content generation, material management, task scheduling, review workflow, and publishing into a single php/postgresql pipeline via rest api, cli, and docker compose.

brreg provides access to norways official business registry via open api, enabling lookups across 1.2 million registered companies by name, number, location, or industry status.

capso-screenshot-macos documents installation and modular architecture of an open-source swift-based screenshot and screen recording tool, but lacks actionable procedures for common developer workflows.

understand-anything-knowledge-graph uses multi-agent analysis to build interactive visual knowledge graphs from codebases, enabling non-linear exploration through a react dashboard with plain-english summaries.

ntwarden-windows-analysis-toolkit provides gui-based inspection of windows processes, kernel internals, services, network activity and etw data locally or remotely via a kernel driver and headless server component.

html-ppt-skill provides static html/css/js presentation templates with 36 themes and 31 layouts, requiring no build tooling. actual procedural depth and failure handling are absent from the documentation.

openclaw-rl-training integrates asynchronous reinforcement learning with live conversation interception to optimize ai agent policies without blocking user interactions.

claude-peers-mcp enables local inter-process communication between multiple Claude Code instances via an MCP server and SQLite broker. message routing and discovery are handled transparently, though setup and trigger guidance remain incomplete.

marketingskills-ai-agents packages marketing frameworks (aida, pas, jtbd) as markdown skills for coding agents, but lacks concrete procedures, specific trigger phrases, and failure-mode guidance needed for reliable execution.

shannon-ai-pentester claims autonomous exploitation of web application vulnerabilities via source code analysis and concurrent exploit agents, but documentation lacks executable procedures and omits critical operational constraints.

multi-market stock analysis that collects quotes and news, generates ai dashboards with buy/sell signals, and distributes results across messaging platforms on a schedule. incomplete operational docs.

awesome-free-llm-apis catalogs permanent free-tier llm providers with rate limits and regional availability. covers openai-compatible endpoints and notable model lineups but lacks integration procedures and failure mode guidance.

weixin-agent-sdk provides typescript bindings for wechat agent integration via long-polling, targeting developers building chatbot backends without public server infrastructure.

voicebox-voice-synthesis is a local-first voice cloning and tts studio exposing a rest api. skill covers installation paths but lacks agent integration points, callable intents, and failure recovery logic.

polymarket-arbitrage-trading-bot executes automated dump-and-hedge trades on polymarket 15-minute crypto markets by detecting mispriced legs and locking structural edges before resolution.

palantir-for-family-trips is a react-based dark dashboard for multi-family road trip coordination, offering convoy tracking, timeline playback, meal logistics, and expense splits via google maps integration.

witr-process-inspector is a go cli tool that traces process ancestry and startup causality across shells, containers, systemd, and supervisors without requiring manual correlation of multiple diagnostic commands.

crucix-intelligence-dashboard aggregates 27 open-source intelligence feeds (satellite, flight, radiation, maritime, market data) into a self-hosted webgl globe with optional llm-enriched alerts to telegram or discord.

open-multi-agent-orchestration covers typescript-based multi-agent team coordination with dag-based task scheduling and in-process execution, but lacks concrete procedure steps and failure handling guidance.

opencli-web-automation bridges websites and cli through chrome sessions, supporting 80+ commands across 19 sites with extensible adapter patterns, but lacks procedural depth and failure handling details.

json-render enables ai agents to generate dynamic ui layouts from natural language by constraining output to a predefined component catalog, with platform-agnostic json rendering and multi-framework support.

bazi-skill-chinese-astrology collects birth data interactively and generates four-pillars destiny charts with cycle analysis and pattern classification, citing nine classical texts. execution quality and error handling remain undocumented.

chrome extension automating openai oauth flows with email verification and multi-round batch support across duckduckgo, qq mail, 163 mail, and inbucket. lacks operational depth and failure recovery detail.

apache-airflow-orchestration covers dag definition, workflow scheduling, and monitoring basics. content is introductory and installation-focused, lacking actionable procedures for real pipeline scenarios.

free-code-claude-cli wraps a telemetry-stripped, guardrail-removed fork of claude code cli with 54 unlocked experimental features, supporting five model providers.

ministack-aws-emulator provides local aws service simulation across 25+ services including s3, sqs, dynamodb, and lambda, running on a single configurable port without credentials or external dependencies.

mole-mac-cleaner wraps a macOS maintenance CLI for disk cleanup and system optimization. the skill documentation provides installation paths but lacks actionable agent procedures, error handling, or integration contracts.

anything-analyzer-cdp is an electron desktop app that intercepts browser network traffic via chrome devtools protocol and sends captured data to ai models for protocol analysis report generation.

tavily-key-generator-proxy automates bulk account creation and pools api keys behind a round-robin proxy gateway. lacks operational depth on failure modes, scaling limits, and deployment specifics.

posterskill-academic-posters generates print-ready conference posters from overleaf papers, outputting a self-contained html file with drag-and-drop editing. setup involves git clones and optional reference pdfs.

minecraftconsoles-lce is a c++ reimplementation of minecraft legacy console edition targeting windows with lan multiplayer and dedicated server support, but the documentation conflates a game engine project with an ai skill.

mm2-roblox-analytics-tracker claims to deliver inventory tracking and performance analytics for murder mystery 2, but lacks procedural depth, trigger specificity, and concrete implementation details.

nvidia-nemoclaw is an alpha typescript cli plugin for running openclaw ai assistants with policy-enforced sandboxes and nvidia cloud inference routing. the documentation is incomplete and lacks actionable procedures.

mcp-brasil wraps 28 brazilian public apis as an mcp server with 213 tools and 55 resources. installation is straightforward via pip or uv, but lacks actionable procedures, failure modes, and domain-specific trigger guidance.

moss-tts-nano-speech enables multilingual real-time speech synthesis via a lightweight 0.1b parameter model supporting voice cloning and cpu-only inference. setup coverage incomplete.

awesome-phd-cv packages three latex resume templates (academic, ats-optimized, two-column) for researchers entering or in industry. no execution logic or decision framework.

open-agent-sdk documents a typescript framework for deploying autonomous agents in-process. installation and basic setup are sketched but procedure depth, trigger examples, and failure handling are largely absent.

everything-claude-code-harness claims to be a performance optimization and security system for ai coding agents across multiple platforms, but the provided skill.md contains only marketing copy, installation instructions, and no actionable procedures.

claude-hud-statusline is a terminal plugin displaying context usage, active tools, running agents, and task progress in a persistent statusline, but lacks actionable setup instructions and decision logic.

copaw-ai-assistant is a multi-platform deployment framework for running a personal ai agent across chat services, local/cloud llms, and custom python extensions via web console and cli.

applies pattern matching between trump social media activity and equity market movements, offering statistical correlation analysis via brute-force model selection.

aris-autonomous-ml-research proposes a markdown-based framework for orchestrating ml research workflows across multiple language models, emphasizing adversarial review loops. lacks concrete procedures and implementation specifics.

isolet-widget-isolation packages components into shadow-DOM-isolated widgets with multi-framework support and multiple output formats, but the skill lacks procedural depth, explicit trigger scenarios, and failure-mode guidance.

openhanako-personal-ai-agent is an electron-based desktop platform enabling persistent multi-agent collaboration with file system, terminal, and web capabilities, but lacks actionable skill integration details.

ui-ux-pro-max-skill claims to inject design reasoning into agents via 161 rules and 67 styles, but lacks actionable procedures, concrete trigger definitions, and failure mode documentation. the skill.md reads as marketing copy rather than executable specification.

murder-mystery-2-analytics-toolkit appears to be a roblox game data aggregation library with inventory tracking and strategy analysis, but lacks clear intent specification, decision logic, and outcome signals required for skill orchestration.

openmaic-classroom is a next.js / react classroom platform that generates interactive lessons from topics or documents, but lacks actionable skill definition, trigger patterns, and procedural detail needed for agent integration.

ghostling-libghostty-terminal is a minimal terminal emulator demonstrating libghostty-vt library integration with raylib, covering vt parsing, terminal state, and rendering primitives in a single-file c codebase.

wterm provides DOM-based terminal emulation with zig/wasm parsing and react bindings, but documentation is incomplete - no usage procedures, error handling, or deployment specifics.

hy-world-2-0-3d-world-model provides access to tencent's multi-modal world model for 3d reconstruction and generation from text, images, and video, but lacks operational guidance and error handling.

fabro-workflow-factory is an orchestrator framework for defining agent pipelines as directed graphs, supporting branching, loops, and human gates, but lacks concrete skill integration guidance.

nanochat-llm-training wraps karpathy's minimal llm harness for single-gpu end-to-end training, covering tokenization through inference, but the skill.md provides only installation scaffolding without actionable training procedures.

toolkit for roblox murder mystery 2 players to track inventory, analyze gameplay statistics, and generate performance reports through command-line and python api interfaces.

grimmory-self-hosted-library covers docker-based book collection management with epub/pdf/comic support, but the skill document is incomplete and lacks actionable setup procedures.

maoxuan-skill-cognitive-framework packages mao zedong's analytical methods as operational mental models for startups, team dynamics, and resource allocation, but lacks procedural specificity and actionable decision trees.

mm2-roblox-analytics-toolkit claims to provide murder mystery 2 game analytics and inventory tracking, but the skill.md cuts off mid-setup without defining intents, inputs, outputs, or actual workflow beyond installation commands.

slug-font-rendering is a reference implementation of gpu-accelerated vector font rendering using bézier curve encoding and fragment shader coverage resolution, but lacks actionable integration guidance for ai agents.

freecodecamp-curriculum documents the open-source platform architecture without actionable procedures, trigger conditions, or recovery guidance. the document fragment provides scaffolding but lacks executable substance.

clawgod-claude-code-patch claims to patch claude code at runtime to unlock hidden features and remove safety restrictions, but provides only installation commands and marketing copy without actionable skill procedures or decision logic.

obliteratus-abliteration packages mechanistic interpretability techniques to locate and remove safety-aligned refusal behaviors from llm weights via svd/pca projection, shipping gradio ui and cli interfaces.

Analyze Murder Mystery 2 gameplay data, track inventory, and optimize strategy using this Roblox analytics toolkit

Multi-platform hot search crawler and LLM-powered public opinion analysis system with clustering, sentiment analysis, and multi-channel push notifications

A collection of specialized AI agent personalities for Claude Code, Cursor, Aider, Windsurf, and other AI coding tools — covering engineering, design,…

Roblox Murder Mystery 2 analytics dashboard and inventory tracking toolkit with data visualization and strategy analysis

Murder Mystery 2 inventory tracking, analytics dashboard, and gameplay optimization toolkit for Roblox

Fully autonomous research pipeline that turns a topic idea into a complete academic paper with real citations, experiments, and conference-ready LaTeX.

flash-moe-inference — an installable skill for AI agents, published by aradotso/trending-skills.

modly-image-to-3d — an installable skill for AI agents, published by aradotso/trending-skills.

git-city-3d-github-visualization — an installable skill for AI agents, published by aradotso/trending-skills.

keyid-agent-kit-mcp — an installable skill for AI agents, published by aradotso/trending-skills.

nightingale-karaoke — an installable skill for AI agents, published by aradotso/trending-skills.

autoresearch-genealogy — an installable skill for AI agents, published by aradotso/trending-skills.

kimodo-motion-diffusion — an installable skill for AI agents, published by aradotso/trending-skills.

clui-cc-claude-overlay — an installable skill for AI agents, published by aradotso/trending-skills.

taiwan-md-knowledge-base — an installable skill for AI agents, published by aradotso/trending-skills.

corridorkey-green-screen — an installable skill for AI agents, published by aradotso/trending-skills.

zeroboot-vm-sandbox — an installable skill for AI agents, published by aradotso/trending-skills.

mirofish-offline-simulation — an installable skill for AI agents, published by aradotso/trending-skills.

toon-format — an installable skill for AI agents, published by aradotso/trending-skills.

karpathy-jobs-bls-visualizer — an installable skill for AI agents, published by aradotso/trending-skills.

open-autoglm-phone-agent — an installable skill for AI agents, published by aradotso/trending-skills.

worldmonitor-intelligence-dashboard — an installable skill for AI agents, published by aradotso/trending-skills.

openai-symphony-autonomous-agents — an installable skill for AI agents, published by aradotso/trending-skills.

cmux-terminal-multiplexer — an installable skill for AI agents, published by aradotso/trending-skills.

metaclaw-evolving-agent — an installable skill for AI agents, published by aradotso/trending-skills.

openclaw-config — an installable skill for AI agents, published by aradotso/trending-skills.

paperclip-ai-orchestration — an installable skill for AI agents, published by aradotso/trending-skills.

gsd-2-agent-framework — an installable skill for AI agents, published by aradotso/trending-skills.

Expert skill for Xata open-source cloud-native Postgres platform with copy-on-write branching, scale-to-zero, and Kubernetes deployment

Python port of Claude Code agent harness — tools, commands, task orchestration, and CLI entrypoint via oh-my-codex

AI-powered browser extension that filters unwanted posts from Twitter/X feeds using natural language rules and multiple AI backends

Sync, search, classify, and query X/Twitter bookmarks locally using the Field Theory CLI

Minimal terminal spinner library for Rust with preset collection and no-std support

Use Claude Code's full tool system with any OpenAI-compatible LLM — GPT-4o, DeepSeek, Gemini, Ollama, and 200+ models via environment variable configuration.

Android app that detects VPN/proxy servers (VLESS/xray/sing-box) via local SOCKS5 vulnerability, exposing exit IPs and server configs without root

Convert files, URLs, and media to markdown using the markit-ai CLI and SDK with pluggable converters and LLM support.

Free open-source macOS cleaner built with SwiftUI — CleanMyMac alternative with zero telemetry, scheduled auto-cleaning, and Xcode/Homebrew/system cache…

PokeClaw (PocketClaw) — on-device Android AI phone agent using Gemma 4 via LiteRT-LM with tool calling, accessibility automation, and optional cloud models.

AI coding agent skill for AlayaRenderer — a generative world rendering framework with inverse rendering (RGB→G-buffers) and game editing…

Full-pipeline AI skill for WeChat Official Account articles — hotspot fetching, topic selection, writing, SEO, image generation, formatting, and draft box…

AI 에이전트를 위한 한국 서비스 자동화 스킬 모음 — SRT/KTX 예매, KBO, 로또, 카카오톡, 지하철, HWP, 우편번호 등

AI-first knowledge base and startup OS with file-based storage, AI agents, scheduled jobs, and embedded apps

童锦程视角 Skill — 用深情祖师爷的思维框架分析人际关系、恋爱问题与人性洞察

Build and extend Gemma Gem, an on-device AI browser assistant Chrome extension running Google's Gemma 4 model via WebGPU with no cloud dependencies.

WildWorld large-scale action-conditioned world modeling dataset with 108M+ frames from a photorealistic ARPG game, featuring per-frame annotations, 450+…

Blazing fast Rust CLI tool to fetch information from 55+ websites (Twitter, Reddit, YouTube, Bilibili, Zhihu, etc.) with a single command, plus AI-native…

AI-native open-source Figma alternative with CLI, MCP server, and Vue SDK for reading/writing .fig files programmatically.

童锦程视角 Skill — 用深情祖师爷的思维框架分析人际关系、恋爱问题与人性洞察

Agent skill to convert any arxiv paper into a citation-anchored, working Python implementation with ambiguity auditing

Interactive TUI dashboard for visualizing Claude Code token usage, costs, and task breakdowns by project, model, and activity type.

Expert skill for using RX, an embedded data store for JSON-shaped data with random-access reads, no-parse lookups, and a text-safe binary encoding format.

ACG Type Indicator — MBTI-inspired anime character persona quiz built with Vue 3, TypeScript, and Vite

Local drop-in API emulation for Vercel, GitHub, and Google services in CI and no-network sandboxes

Structured equity research database for 1,735 Taiwan-listed companies with wikilink knowledge graph, supply chain mapping, and financial data tools.

Drop-in SwiftUI sticky header component with progressive blur effect matching Apple Music, Photos, and App Store styles.

Feed-forward 3D foundation model for streaming scene reconstruction using Geometric Context Transformer

AI-powered deep stock analysis engine for A-share/HK/US markets with 51 investor personas, 22 data dimensions, 180 quantitative rules, and 17 institutional…

WebGPU rigid-body/soft-body physics engine based on the AVBD (Augmented Vertex Block Descent) solver

Build and maintain a personal knowledge base using Karpathy's llm-wiki methodology across Claude Code, Codex, and OpenClaw agents.

Create accurate Japanese UI DESIGN.md files for AI agents with proper CJK typography, font stacks, line-height, kinsoku shori, and mixed typesetting rules.

Deep dive into Claude Code source code to learn production-grade AI agent architecture patterns

tailslayer-dram-hedged-reads — an installable skill for AI agents, published by aradotso/trending-skills.

Run and interact with KarpathyTalk, an open markdown-based developer social network with GitHub auth, SQLite, and an LLM-friendly JSON/markdown API.

AI-powered penetration testing assistant using local LLM (metatron-qwen via Ollama) on Parrot OS Linux