Anduin2017/HowToCook 是一个由社区驱动的中文菜谱仓库，旨在为程序员提供形式化、可量化的烹饪指南。该仓库包含超过 700 个 Markdown 菜谱文件，按照菜系（素菜、荤菜、水产、早餐、主食等）组织，每个菜谱都遵循严格的模板格式，包含精确的计量、步骤和卡路里预估。仓库还包含辅助工具如自动 README 生成器（readme-generate.js）、Markdown 质量检查脚本（manual_lint.js）以及 GitHub Actions CI 流程，确保菜谱的一致性和可读性。虽然本质上是一个文档/数据仓库，但其组织方式、质量控制工具和社区协作模式使其成为一个高质量的开源数据项目，方便被其他应用（如菜谱API、小程序）消费。
This repository serves as the primary source for .NET documentation (learn.microsoft.com/dotnet). It contains conceptual guides, API references, and runnable code samples demonstrating .NET language features, frameworks, and best practices. The repository is not a single application but a curated collection of educational materials organized by technology domain (e.g., C#, ASP.NET Core, Orleans, AI). Code samples illustrate patterns like dependency injection, logging, serialization, and data access, often with minimal, self-contained examples. Maintenance is active via GitHub Actions workflows for linting, dependency updates, and version sweeping, ensuring content stays current with .NET releases.
This repository is a curated documentation project, not a software application or a traditional data pipeline. It functions as a 'Wall of Shame' for mobile device manufacturers and carriers that restrict bootloader unlocking. The 'data' is the collective knowledge of the community, structured into Markdown files with a clear taxonomy (verdicts: Avoid, Terrible, Caution, Safe). The primary data flows are manual curation and community contributions guided by strict formatting rules. It answers the research question: 'Which companies prevent me from controlling my device, and how can I bypass their restrictions?' The repository demonstrates strong data engineering principles in its organization. The directory structure (`brands/<company>/README.md`, `carriers/brands/<carrier>/README.md`) acts as a schema, and the `TEMPLATE.md` and `CONTRIBUTING.md` files define a rigorous data ingestion and validation process. The content itself is highly detailed, with deep dives into specific exploits, legal agreements, and hardware-level bypasses, making it a definitive reference for the mobile device modding community. While it lacks executable code, ETL scripts, or machine learning models, its value as a research artifact is exceptionally high. The reproducibility is perfect: an analyst can clone the repository and immediately access the full dataset without any dependencies or environment setup. The project is actively maintained, with recent commits and a responsive issue tracker.
The ai-boost/awesome-prompts repository is a curated collection of prompt templates designed to configure AI agents for specific roles and tasks. It contains hundreds of specialized prompts for agent architectures (e.g., personal agent brains, multi-agent orchestrators), agent skills (e.g., coding agents, security auditors), and domain-specific agents (e.g., healthcare operations, legal analysts). Each prompt encapsulates non-obvious design patterns for agent systems, such as three-layer retrieval for memory, plan-execute loops, declarative UIs, and skill-based architectures. The repository serves as a reference for prompt engineering and agent design, enabling users to build reliable, safe, and effective LLM-based agents without starting from scratch.
This repository is a curated directory of internet radio streams, organized by genre, with automated tooling to generate M3U playlists, check link health, and probe stream availability. It is primarily a data curation project, with the README.md serving as the source of truth. The supporting scripts form a robust data pipeline: parsing the markdown, generating playlists, and running health checks with retry logic and silence detection. The project demonstrates strong data engineering practices in a non-traditional data context, with shared parsers, persistent state tracking, and CI workflows that handle race conditions gracefully.
This repository is a community-maintained, curated registry of EVM-compatible blockchain networks. It contains thousands of JSON files under `_data/chains/`, each defining a chain's metadata: RPC endpoints, native currency, explorers, icons (IPFS CIDs), status (active/deprecated/incubating), parent L2 relationships, and feature flags. The dataset is the primary asset, designed for consumption by wallets, dApps, block explorers, and infrastructure tooling that needs up-to-date chain information. Beyond the data, the repository includes a Kotlin-based processor that validates all chain definitions (schema compliance, icon size/format, IPFS retrievability, ENS address checks) and produces aggregated `chains.json` and `chains_mini.json` outputs. A published Kotlin model library (`model/`) and an HTTP loader (`httpsloader/`) enable programmatic access. CI pipelines enforce formatting, schema validation, and per-PR checks. The project is hybrid: its core value is the dataset, but it also delivers a well-engineered application layer for validation, aggregation, and distribution. The registry answers questions like 'What RPC endpoints are available for chain X?', 'What is the native currency of chain Y?', 'Which chains are L2s of Ethereum?', and 'What explorers support EIP3091?'. It serves as a single source of truth for EVM chain metadata, reducing fragmentation across the ecosystem.
This repository is a comprehensive, practitioner-focused educational hub for generative AI. It is not a deployable software application or a traditional data pipeline, but a curated knowledge base that bridges theory and practice. The content is organized into free courses (Applied LLMs Mastery, Agentic AI, AI Evals), interview preparation guides for multiple AI roles (AI Engineer, AI PM, FDE, AI Strategist), system design case studies with runnable LangGraph code, and curated research updates (monthly paper lists, State of AI reports). The architecture is a hybrid of educational content and reusable code patterns, with the core value lying in its structured learning paths and conceptual frameworks. The repository's dominant character is educational content curation and research tracking. The code examples are explicitly demonstration-grade, designed to teach production-adjacent patterns like provider-agnostic LLM layers, hash-chained audit trails, and deterministic guardrails. These patterns are genuinely reusable and demonstrate strong engineering craftsmanship. The information architecture of the repo itself is excellent, making it easy for learners to navigate from foundational concepts to advanced system design. The repository is actively maintained with a massive community following. Its primary risk is the inherent rapid obsolescence of the AI field, which the maintainers mitigate by explicitly archiving outdated content and providing clear versioning. The code examples are not production-ready out of the box, but the patterns they teach are directly applicable to real-world systems. The research paper lists and interview questions require ongoing curation to stay current.
global-stock-data is a comprehensive financial data pipeline designed for AI coding assistants to access US and Hong Kong market data. It implements a 13-layer architecture that fetches data from 11 diverse sources including exchanges (CBOE), regulatory bodies (SEC, FINRA), financial data providers (Eastmoney, Yahoo Finance), and macroeconomic sources (Treasury, CFTC). The system normalizes data from various formats (JSON, CSV, text) into consistent schemas for real-time quotes, historical prices, technical indicators, fundamentals, options data, SEC filings, short volume, and macroeconomic indicators. Unlike a traditional application, it functions as a reusable data toolkit where each layer can be independently accessed to solve specific financial data acquisition problems.
This repository is the official companion code for the book "Build a Reasoning Model (From Scratch)" by Sebastian Raschka (Manning, 2025). It is primarily a data_research codebase focused on building, evaluating, and training reasoning LLMs from scratch using the Qwen3 0.6B model as a base. The code covers inference-time scaling (temperature/top-p sampling, self-consistency), reinforcement learning from verifiable rewards (GRPO/RLVR), self-refinement, and knowledge distillation from larger teachers (DeepSeek-R1, Qwen3 235B). It includes a full custom Qwen3 implementation with KV cache, RoPE, and grouped-query attention, plus training scripts, notebooks, benchmark evaluations (MATH-500, MMLU), and Hugging Face export adapters. While educational in nature, it demonstrates production-quality engineering for research experiments. The repository is hybrid: it provides a reusable Python package for text generation and training, but its dominant character is data_research due to its notebook-driven, experimental, and benchmark-focused workflow.
This repository is a comprehensive architectural analysis of the leaked Claude Code CLI source code, presented as 18 detailed markdown documents. It systematically dissects the entire 512K-line TypeScript codebase across all subsystems: startup pipeline, query engine with AsyncGenerator state machine, 39-tool system with 30+ method interfaces, three-layer Bash security (23 checks), ResolveOnce permission system with 6-way concurrent race handling, 150+ field immutable state management, multi-provider API service layer, dual-transport IDE Bridge integration, file-based memory system with Sonnet semantic recall, lazy-loaded plugin system, custom React reconciler with double-buffered terminal rendering, virtual scrolling with Float64Array offset tracking, 103-command system, build-time DCE with GrowthBook runtime gates, and 12 cross-cutting architectural patterns. While the repository does not contain the actual leaked source code, it serves as an exceptionally well-structured reference for understanding how a production-grade AI coding assistant is architected. The analysis includes code snippets, design rationale, architecture diagrams, and performance optimization details. The work is of publication quality, demonstrating deep software engineering expertise and systematic analytical methodology. Researchers and engineers studying AI-assisted programming tools, terminal UI frameworks, or industrial TypeScript architectures will find this a valuable reference.
Stable-worldmodel is a research framework for goal-conditioned world models in reinforcement learning, providing a full pipeline from data collection and storage to model training and planning-based evaluation. The codebase is primarily a research platform with strong software engineering practices—pluggable data formats (Lance, HDF5, video, folder), a command-line interface (swm), and a Hydra-based configuration system for reproducible experiments. It includes implementations of multiple world model architectures (LeWM, PLDM, PreJEPA/DINO-WM, TD-MPC2, GCRL), planning solvers (CEM, MPPI, iCEM, gradient-based, Lagrangian), and environment wrappers with domain randomization across 30+ tasks (DeepMind Control, Gymnasium, PushT, OGBench, Craftax, robotics). The repository demonstrates high craftsmanship in its data engineering: the data format registry extends cleanly, the Lance format supports cloud-native streaming from Hugging Face Buckets, and utilities like GoalDataset, HistoryBuffer, and ReplayBuffer solve recurring RL data management problems. The project is actively maintained with extensive test coverage, comprehensive documentation, and a scalable training pipeline using stable-pretraining and Lightning. While the primary goal is research (testing hypotheses about world model generalization under variation), the codebase includes application-level production quality—CI/CD, CLI, checkpoint management, and robust error handling—making it a rare hybrid that serves both researchers and engineers.
The Data Engineering Wiki is a community-maintained knowledge base dedicated to documenting data engineering concepts, tools, and best practices. Rather than being an executable application or data processing pipeline, it serves as a comprehensive reference resource for individuals learning and working in the data engineering field. The repository is organized into thematic sections covering foundational concepts (data architecture, modeling, processing, storage), practical tools and technologies, implementation guides, frequently asked questions, and curated learning resources. Structurally, the wiki follows a logical progression from theoretical foundations to practical application, with concepts explained through clear definitions, diagrams, and real-world examples. Key strengths include its comprehensive coverage of both established and emerging data engineering topics, practical guides on data pipeline best practices and testing methodologies, and active maintenance evidenced by recent commits. The content is presented in an accessible markdown format with consistent frontmatter, making it easy to navigate and consume. While not containing executable code or data pipelines itself, the wiki provides reusable knowledge assets in the form of methodological guides and reference materials that can be directly applied to data engineering projects. Notably, the Data Pipeline Best Practices and Testing Your Data Pipeline guides offer actionable frameworks for building reliable, scalable data systems. The repository's value lies in its role as a living document that captures and organizes community knowledge to support learning and professional practice in data engineering.
This repository, nuwa-skill, is a data_research project that functions as a cognitive framework extraction engine. Its primary output is a collection of structured, testable AI 'perspective skills' (SKILL.md files) that model the thinking patterns, decision heuristics, and expression DNA of prominent public figures (e.g., Elon Musk, Richard Feynman, Steve Jobs, Donald Trump). The system does not run a live service; instead, it provides a rigorous methodology and a set of reusable artifacts for transforming unstructured biographical research into executable AI personas. The core value lies in its extraction framework, fidelity scorecard, and standardized skill template, which together form a complete pipeline for persona distillation. The architecture is a hybrid of a research methodology and a data artifact repository. The data flow begins with a 6-dimension research pipeline (writings, conversations, expression DNA, external views, decisions, timeline) that collects and structures raw source material. This is followed by the application of the Extraction Framework, which uses triple verification to identify mental models and quantify expression DNA. The output is validated against the Fidelity Scorecard through independent dual-agent testing. Supporting scripts (quality_check.py, merge_research.py, srt_to_transcript.py) automate parts of the validation and data processing pipeline, while the CI system (community-pr-check.yml) enforces structural quality gates for community contributions. The most reusable components are the methodology documents (extraction-framework.md, fidelity-scorecard.md, skill-template.md), which solve the recurring data engineering problem of converting unstructured text about a person into a structured, testable cognitive model. The project is actively maintained with a clear contribution model and low technical drift, as its core value is in its markdown-based methodology rather than complex software dependencies. However, reproducibility for creating *new* skills from scratch is low due to the lack of a formalized, dependency-managed environment, making the existing bundled skills the primary reusable asset.
This is the ZJU ICICLES (浙江大学课程攻略共享计划) repository - a massive, student-maintained collection of course materials from Zhejiang University spanning 20+ academic disciplines. The repository contains lecture slides (PPT/PDF), lecture notes (Markdown/LaTeX/Typst), exam papers with solutions, homework assignments with code implementations, lab reports, textbook PDFs, and study guides. While classified as 'data_research' due to the presence of code artifacts in lab directories (compiler construction, data structures, operating systems, computer networks, digital logic), the dominant content is educational courseware - not a data pipeline or ML project. The repository serves as a knowledge base for ZJU students and demonstrates exceptional community curation. Key technical components include: a complete C compiler implementation (lex/yacc + codegen) in 编译原理/实验/, data structure implementations (red-black trees, graph algorithms, B-trees) in 数据结构基础/作业/dsHomework/, RISC-V OS lab code in 操作系统/ and 计算机组成与系统结构/, socket programming labs in 计算机网络/Lab/, and Verilog/FPGA code in 逻辑与计算机设计基础/ and 汇编与接口/. The repository is actively maintained with recent commits, but reproducibility of code artifacts is low due to course-specific dependencies, missing build configurations, and environment assumptions.
This repository is a comprehensive educational and research resource for low-level systems programming, focusing on Linux kernel modules, baremetal programming, CPU architectures (x86_64, ARM, AArch64), and emulation with QEMU and gem5. It contains hundreds of self-contained examples—from trivial 'hello world' kernel modules to advanced topics like IRQ handling, RCU stalls, inline assembly patterns, CPU pipeline hazards, and floating-point arithmetic. The repository is not a production application but rather a knowledge base with automated build and test infrastructure that systematically verifies correctness across multiple architectures and emulators. It also serves as a reference for gem5 simulation, including checkpoint/restore, performance counters, and benchmarking (e.g., Dhrystone, cache performance). While primarily a research/teaching tool, it includes a reusable Python automation framework (CLI generators, parallel test runners, path-based property configuration) that could be extracted for other projects. The repository is hybrid in nature: its core content is data/research (experimental, analytical), but it incorporates application-quality infrastructure.
This repository implements generative recommendation models, primarily HSTU (Hierarchical Sequential Transduction Units) and DLRM-v3, for large-scale sequential recommendation. It is a hybrid codebase combining research experimentation (ICML'24 paper on generative recommenders) with production-grade inference benchmarking (MLPerf loadgen integration). The code provides custom high-performance CUDA and Triton kernels for flash attention, jagged tensor operations, and layer normalization, targeting modern NVIDIA GPUs (Hopper/Blackwell). It includes training pipelines with gin configuration for datasets like MovieLens, Amazon Books, KuaiRand, and synthetic streaming data up to 100B items. The repository also contains a full MLPerf inference benchmark suite with vendored loadgen library and end-to-end TorchScript tracing for deployment. While the core focus is research on scaling laws and architecture innovations for sequential recommenders, the ops and modules are engineered to production standards with multiple kernel backends and rigorous tests.
This repository is a comprehensive French-language training course for PostgreSQL 18, structured as 22 progressive chapters and 7 reference annexes. It serves as educational material covering PostgreSQL fundamentals, SQL querying, advanced features, administration, performance tuning, extensions, and modern architectures. The content combines theoretical explanations with practical examples, best practices, and version-specific features (like asynchronous I/O, UUIDv7, and native OAuth). While not containing executable code or data pipelines, it provides reusable reference materials such as SQL query templates, shell scripts for backup/monitoring, configuration guides, and checklists that can be adapted across projects. The repository is actively maintained with recent updates and demonstrates low technical drift due to its focus on enduring database principles alongside version-specific annotations.
This repository is a distribution of the classic 'employees' sample database, a realistic relational dataset of ~300,000 employees, ~2.8 million salary records, and associated department/title/manager data. It is not an application but a data asset and benchmark fixture used to test database servers, application integrations, and migration tooling. The repo contains MySQL and PostgreSQL schemas, raw data dumps, partitioned table variants, stored procedures/views, and an integrated integrity-test suite that verifies row counts and incremental MD5/SHA-1/SHA-256 checksums. It also bundles a version of the Sakila sample database. CI workflows exercise loading and integrity checks across MySQL 5.6-9.6, Percona Server, MariaDB, and PostgreSQL 16/17, and the SHA-256 checksums are intentionally identical across all supported engines. The primary value is a non-trivial, self-contained dataset with strong cross-database compatibility and a clever, reusable data-integrity verification pattern.
This repository is a meticulously curated knowledge base and toolkit for AI video prompt engineering. Its core is a structured data pipeline: raw knowledge from official vendor documentation and YouTube masterclasses is reverse-engineered into standardized methodologies (21+ SOPs) and a centralized prompt database (`prompts/data/all-prompts.json`). This data is then exposed through a cross-model benchmark (`prompts/data/cross-model-matrix.json`), zero-dependency web tools (Prompt Browser, Cross-Model Matrix), and executable Claude Code Skills (`skills/`). The architecture is data-centric. The primary value lies in the curated, structured data assets and the rigorous methodology used to create them. The application layer (web tools, skills) is a thin, well-crafted consumer of this data. The repository actively manages its primary risk—model version drift—through an automated monitoring system (`scripts/monitor_models.py`) that hashes external endpoints and files issues on change. While primarily a data_research project, it exhibits strong application characteristics through its interactive tools and AI skill integrations. It is a living knowledge base, actively maintained with a clear contribution workflow and a strong emphasis on data provenance and reproducibility. The `cross-model-matrix.json` is a standout asset, providing a controlled benchmark for comparing prompt styles across 11 major AI video models.
This repository is a comprehensive, curated bibliography and knowledge base for the field of active learning. It organizes hundreds of research papers by conference (ICML, NeurIPS, CVPR, etc.), journal (JMLR, TPAMI, etc.), thesis, and content topic (problem settings, deep active learning, applications, etc.). The repository also provides taxonomies, summaries, and links to papers. It is not a software application or data pipeline; it is a structured reference list intended for researchers and practitioners. The content is entirely in Markdown files, making it trivially reproducible and easy to navigate. The repository is actively maintained with recent commits and a growing star count.