Context7 integrates current, version-specific documentation and code examples directly into your LLM's context, eliminating outdated information and hallucinations. It's ideal for developers needing precise, up-to-date API details, configuration, or code snippets for any library or framework.

The Claude Cookbooks provide practical, copy-paste code snippets and guides to help developers integrate Claude's AI capabilities into their projects. It's an essential resource for exploring various use cases, from basic text processing to advanced agent orchestration and multimodal applications.

GitNexus transforms any codebase into a detailed knowledge graph, mapping dependencies, call chains, and execution flows. This provides AI agents with deep architectural awareness, significantly improving their reliability for complex development tasks. It is particularly effective for developers using AI assistants to navigate, debug, and refactor large codebases with precision.

This skill provides an AI agent-led search engine that aggregates and synthesizes recent public engagement from diverse platforms like Reddit, X, YouTube, TikTok, GitHub, and Polymarket. It excels at revealing real-time trends and public sentiment, making it ideal for market research, competitive analysis, or preparing for critical meetings.

This repository offers 135 ready-to-use scientific and research skills, transforming any AI agent supporting the Agent Skills standard into a powerful research assistant. It excels in automating complex multi-step scientific workflows across diverse domains like bioinformatics, drug discovery, clinical research, and materials science, integrating with over 100 scientific databases and optimized Python packages.

This collection offers a comprehensive set of Agent Skills focused on context engineering principles for building and optimizing production-grade AI agent systems. It provides structured guidance and practical examples for managing the language model's context window effectively across various agent platforms.

This repository provides a collection of standardized Agent Skills for AI/ML tasks, including model training, evaluation, and dataset management on the Hugging Face Hub. It is designed for developers using coding agents like Claude Code, Codex, or Gemini CLI to streamline complex machine learning workflows.

Upsonic is a Python framework designed for building both autonomous and traditional AI agents, offering capabilities to automate complex tasks and integrate with external services. It excels in scenarios requiring robust agent orchestration, custom tool integration, and efficient document processing with built-in OCR.

This skill enables Claude Code to directly query Google NotebookLM, providing intelligent, citation-backed answers exclusively from your uploaded documents. It's ideal for developers needing precise, hallucination-reduced information from their technical documentation without manual copy-pasting.

This Claude Code skill decompiles Android APK, XAPK, JAR, and AAR files, extracting HTTP API endpoints like Retrofit, OkHttp calls, and hardcoded URLs. It's invaluable for security research, malware analysis, or documenting and reproducing app functionalities without access to the original source code.

This skill suite provides a comprehensive, human-in-the-loop pipeline for academic research, covering everything from initial planning and deep research to paper writing, peer review, and final integrity checks. It excels at augmenting human researchers by handling grunt work and verifying data, allowing focus on critical thinking and interpretation.

This Claude Code skill transforms your Obsidian vault into an autonomous knowledge engine that creates, organizes, maintains, and evolves notes. It ingests sources, extracts entities, updates cross-references, and files everything into a structured wiki. It's ideal for building a persistent, compounding knowledge base that answers questions by synthesizing information from your entire vault, citing specific pages.

This MCP server integrates AI assistants with Exa's specialized search engine, providing real-time access to web, code, and company data. It is ideal for AI agents requiring up-to-date external information for research, analysis, or contextual understanding.

This MCP server allows AI assistants to programmatically search, download, and access content from arXiv's research repository. It's ideal for AI agents needing to perform research, analyze scientific papers, or monitor new publications in specific fields.

This Claude Code artifact offers a comprehensive, file-based personal health information system, enabling users to manage medical reports, track health metrics, and receive AI-driven insights. It excels in providing private, local data storage with advanced features like image recognition for reports and multi-specialty consultations, making it ideal for individuals seeking detailed, secure health oversight without cloud dependencies.

This marketplace provides a comprehensive suite of AI agent skills tailored for solopreneurs and indie hackers, enhancing various business operations. It equips AI agents with specialized capabilities for market research, SEO, design, and social media, enabling efficient task execution across numerous AI coding environments.

Octocode is an MCP server that enhances AI assistants with deep code understanding by integrating GitHub, GitLab, and local LSP intelligence. It excels at providing comprehensive context for AI-driven development, research, and complex code analysis tasks.

BlockRun MCP integruje różnorodne źródła danych w czasie rzeczywistym i zaawansowane modele AI bezpośrednio z Claude, umożliwiając mu odpowiadanie na złożone, aktualne pytania dotyczące rynków, badań i mediów społecznościowych, a także generowanie treści multimedialnych. Jest to idealne rozwiązanie dla deweloperów potrzebujących dostępu na żądanie do danych premium i możliwości AI bez zarządzania wieloma kluczami API czy subskrypcjami.

RivalSearchMCP is a FastMCP 3.x server providing 10 specialized tools for comprehensive, auditable research. It integrates 5 web search engines, 9 social platforms, 5 news sources, 5 academic databases, 4 dataset hubs, and GitHub. Features include document analysis (PDF, Word, OCR for images) and website traversal. Crucially, it returns deterministic, structured outputs without an in-server LLM, enabling robust agent chaining. It offers auto-quality scoring, conflict detection across sources, and persistent research workspaces for iterative investigations.

This Claude Skill provides specialized knowledge and workflows for family history and genealogy research. It guides users through creating structured research plans, generates properly formatted genealogical citations following the Evidence Explained methodology, and systematically analyzes conflicting evidence according to the Genealogical Proof Standard (GPS). It also assists in creating professional research logs. The skill emphasizes privacy and responsible AI use, adhering to CRAIGEN principles, and ensures a methodical approach to research, prohibiting unsolicited web searches.

This skill implements Recursive Language Model (RLM) strategies to overcome Claude's context window limitations and 'context rot'. It enables Claude Code to process inputs up to 2 orders of magnitude beyond normal limits by systematically filtering, chunking, recursing with sub-agents, verifying, and synthesizing information. Ideal for analyzing large codebases, multi-document QA, and complex information aggregation, it significantly improves accuracy and completeness on tasks involving 10+ files or over 50k tokens. Benchmarks show substantial improvements in multi-hop QA and quadratic reason

This project provides a collection of open-source Agent Skills designed for computational materials science and numerical simulation. It addresses the problem of general-purpose AI agents lacking domain-specific knowledge, requiring constant hand-holding for tasks like checking CFL numbers or selecting appropriate solvers. By packaging this expertise into structured skills, agents can automatically discover, load, and execute validated Python scripts. This enables AI agents to perform complex scientific computing tasks, from numerical stability analysis and mesh generation to HPC job script g

This skill provides comprehensive Kaggle integration for AI coding agents like Claude Code. It automates account setup, generates competition landscape reports, facilitates dataset and model downloads, executes notebooks, and manages competition submissions. Key features include hackathon writeup retrieval, systematic badge collection, and direct access to 66 Kaggle MCP tools. It emphasizes security with robust credential handling and prompt-injection guards, ensuring safe interaction with Kaggle resources.

TasteRay Skills equip AI agents with the ability to conduct psychological profiling and deliver highly personalized recommendations. It features two main skills: "Elicitation" for understanding user values, motivations, and formative experiences through research-backed conversational techniques (e.g., McAdams' Life Story Interview, Motivational Interviewing), and "Recommendations" which uses this deep understanding to suggest movies, restaurants, products, travel, or jobs via the TasteRay API. This approach moves beyond simple surveys to foster authentic self-disclosure and explain *why* a re

This repository offers a vast collection of over 186 specialized Claude AI skills, meticulously crafted for faculty, researchers, and academicians. Organized across 13 diverse research domains, these skills transform Claude into a domain-specific expert assistant. They are built on real scientific methods, professional output templates, and deep domain knowledge, enabling tasks from multi-agent research pipelines to database integrations and publication-quality output. Users can easily integrate them into Claude Projects or Claude Code CLI for instant expertise in areas like genomics, clinica

A sophisticated research skill for Claude Code that performs deep, multi-cycle investigations across any domain. It uses a Weighted Decision Matrix (WDM) and Munger Inversion to ensure objective analysis, tiers sources (Tier 1-3) to filter out SEO spam, and saves checkpoints to disk to survive context window limits. It integrates with MCPs like Firecrawl and Context7 for high-quality data retrieval, aiming for PhD-level depth through auto-regulated cycles.