SillyTavern is a full-stack web application (Node.js/Express backend, jQuery/Webpack frontend) that serves as a highly customizable frontend for interacting with large language models (LLMs) for roleplaying and chat. It supports dozens of AI backends (OpenAI, Anthropic, Kobold, Ollama, vLLM, etc.) via streaming text and chat completions, and provides an extensive plugin/extension system for TTS, translation, image generation, vector search, and more. The application features a sophisticated macro engine (Chevrotain-based CST parser) for dynamic prompt templating, a slash command framework with typed arguments and autocomplete, and a rich UI with themes, backgrounds, character management, and branching chat history. It is designed for power users who want fine-grained control over prompts, model parameters, and UI behavior, and includes multi-user support with authentication, secrets management, and per-user data isolation.
Crawlkit is a shared Go library and CLI tool (crawlctl) for building local-first crawler archives. It provides provider-neutral building blocks including SQLite stores with FTS5, snapshot/export to gzipped JSONL shards, Git-based mirroring with immutable tags, encrypted backup with age/gzip and SHA-256 deduplication, sync state management with legacy adapters, vector search (exact cosine similarity and optional turbovec Python bridge), embedding providers (OpenAI, Ollama, llama.cpp), a terminal UI (Bubble Tea) for browsing archive rows, and a remote archive client for Cloudflare Workers. The CLI discovers installed crawl apps (e.g., gitcrawl, discrawl) and runs configured refresh jobs under a single-process lock. The architecture is modular and composable, with clear ownership boundaries documented in docs/boundary.md. The repository also includes agent skills (autoreview, crabbox) for automated code review and remote validation, reflecting heavy agent-driven development. It is not a generic web crawler but a mechanics layer that makes each crawl app smaller and more uniform.
This is the official Supabase SSR library (`@supabase/ssr`), a TypeScript package that provides `createBrowserClient` and `createServerClient` functions for using Supabase authentication in server-side rendering frameworks like Next.js, Remix, SvelteKit, and React Router. The library manages user sessions entirely via cookies, handling cookie chunking (splitting large JSON payloads across multiple cookies to respect browser size limits), base64url encoding for safe cookie values, and PKCE code verifier lifecycle. It exposes a `getAll`/`setAll` cookie interface that framework adapters implement to read and write cookies from request/response objects. The server client uses lazy session initialization and writes updated tokens back to cookies on auth state changes, with cache-control headers to prevent CDN caching of auth responses. The package also includes a `clearAuthCookiesAtScopes` migration helper for cleaning up stale cookies after domain/path changes. The codebase is well-tested, actively maintained, and follows modern TypeScript practices with strict mode and no top-level side effects.
This repository is a collection of 100+ Python programming exercises with solutions, organized by difficulty level (Beginner/Level 1, Intermediate/Level 2, Advanced/Level 3). Each exercise includes a question, hints, and a solution. While not a production application, it contains numerous self-contained code snippets that demonstrate practical Python programming concepts and patterns. The exercises cover fundamentals like loops and conditionals, data structures, object-oriented programming, file operations, and advanced topics like generators, decorators, and standard library modules. The primary value lies in its reusable code examples that illustrate specific programming techniques.
Alchemist is a Flutter package designed to simplify and enhance golden testing in Flutter applications. It addresses the common challenge of platform-specific text rendering inconsistencies by providing two types of golden tests: platform tests that render text normally for local development and visual verification, and CI tests that replace text with colored blocks to ensure consistent results across different operating systems and CI environments. The package features a sophisticated configuration system using Dart zones that allows hierarchical configuration management, enabling developers to set global defaults while overriding specific settings for individual tests or test groups. It provides a clean API for defining test scenarios, organizing them in groups, and automatically handling the complexities of widget pumping, image capture, and comparison with configurable tolerance for minor rendering differences. Technical highlights include a custom painting context that replaces text with colored rectangles for CI consistency, a flexible grid layout system for organizing multiple test scenarios, and robust resource management that ensures proper cleanup after each test. The package maintains 100% test coverage and demonstrates strong engineering practices through its use of immutable data structures, proper resource management, and clear separation of concerns.
Kalker is a multi-platform, feature-rich calculator application written primarily in Rust, with web (Svelte + WASM) and mobile (Android via Capacitor) frontends. The core kalk library provides a lexer, parser, AST interpreter, and a numeric value system supporting real, complex, vector, and matrix arithmetic with arbitrary precision (via the rug crate) or standard f64. It includes user-defined variables and functions, calculus operations (differentiation, integration), equation solving via Newton-Raphson inversion, and a comprehensive set of built-in mathematical functions. The CLI offers a REPL with syntax highlighting and tab completion, while the web component is published as an npm package for embedding. The architecture is modular with a clear separation between parsing, evaluation, and frontend, though some code duplication exists between the two numeric backends.
This repository implements a Linux desktop GUI for managing WireGuard VPN profiles. It is built using the Tauri v2 framework, combining a Rust backend for system-level operations (connection management, encryption, file I/O) with a Next.js 16 (React 19) frontend for the user interface. The application allows users to create, import, export, connect, and disconnect WireGuard tunnels, and provides optional at-rest encryption for profile data using a user-defined PIN. The architecture follows a clean IPC-based separation of concerns. The frontend communicates with the Rust backend exclusively through typed Tauri commands defined in `lib/effects.ts`. The backend in `src-tauri/src/main.rs` manages application state, executes WireGuard commands via shell scripts (`wg.sh`), and handles encryption via a dedicated security module. The UI is built from a comprehensive library of ~40+ composable, accessible components in `components/ui/`, demonstrating a strong commitment to frontend architecture and design system principles. The most reusable assets are the Rust security module (`security.rs`), which implements PIN-based encryption using Argon2id and XChaCha20-Poly1305 in a self-contained manner, and the typed IPC service layer (`lib/effects.ts`), which cleanly decouples the UI from the backend. The UI component library, while styled for this project, follows an architectural pattern (atomic, composable primitives) that is directly transferable to other React projects. The project is actively maintained with CI/CD, E2E testing, and semantic versioning, though it carries some technical drift in its Nix build configuration (targeting Tauri v1) and deprecated plugin permissions.
TheAlgorithms/Java is a comprehensive, community-maintained collection of over 600 algorithm and data structure implementations in Java, organized into topic-specific packages (e.g., maths, sorts, searches, dynamic programming, graph, ciphers, compression, machine learning, streaming). It is designed as an educational reference library, not a production application. The architecture follows a modular utility-class pattern: most algorithms are implemented as final classes with private constructors and static methods, and many implement shared interfaces such as SortAlgorithm and SearchAlgorithm for polymorphic usage. The codebase includes extensive JUnit 5 test coverage with parameterized tests, and many classes reference Wikipedia or academic sources. While the overall quality is high with many reusable components, there are some inconsistencies such as duplicate implementations, leftover main methods, and occasional use of raw types or legacy APIs. The repository is actively maintained with CI workflows, dependabot, and a structured contribution process.
This is Microsoft's open-source Windows Calculator, the production calculator app shipped on Windows. The repository is a full UWP/XAML application with Standard, Scientific, Programmer, Date, Unit Converter, and Graphing modes. The core is split into `CalcManager` (a native C++ calculation engine implementing arbitrary-precision rational arithmetic via the Ratpack library), `CalcViewModel` (C++/CX view models handling history, memory, copy/paste, unit data, navigation, and state snapshots), `Calculator` (the C#/XAML UI shell), and `GraphControl` (a DirectX-based graphing control). The app is MVVM-structured, heavily localized in 25+ languages, and includes both C++ unit tests and WinAppDriver-based UI tests.
This repository contains the Apollo Router, a high-performance GraphQL federation gateway written in Rust. In production, it accepts GraphQL queries, mutations, and subscriptions, plans their execution across multiple federated subgraphs, executes subgraph requests, and merges responses. It also supports connectors that integrate REST/HTTP endpoints via @connect/@source directives, enabling a unified GraphQL interface over heterogeneous backends. The router provides extensive observability through OpenTelemetry (metrics, tracing, logging) with exporters for Prometheus, Datadog, Apollo Studio, and OTLP, and includes built-in plugins for authentication (JWT, @authenticated, @requiresScopes), response caching (Redis), demand control, circuit breaking, file uploads, and Rhai scripting for custom logic. The architecture is a monolithic binary with a plugin-based service pipeline built on a custom async runtime (tower). Key pipeline stages include the router service (HTTP handling), supergraph service (query planning), execution service (orchestration), and subgraph/connector services (outbound HTTP). Configuration is YAML-based with hot-reload support. The federation library (apollo-federation) handles subgraph schema merging, composition, query plan generation (including an incremental planner with BULB search), and correctness validation. The codebase is large (4500+ files) with extensive unit, integration, and fuzz testing, and follows a rapid release cadence (every 1-2 weeks). Key features include: a custom JSON selection DSL for connector data transformation, a typed cache tag system for CDN invalidation, an immutable persistent list (SharedPath) for efficient query plan path forking, and a comprehensive telemetry configuration system with selectors and conditions. The router is designed for production-scale workloads and is regularly benchmarked for performance and memory usage.
This repository implements the official GitHub Action `actions/setup-node`, which sets up a Node.js environment in GitHub Actions workflows. It downloads and caches the requested Node.js version (including LTS, nightly, RC, and v8-canary builds), adds it to the PATH, optionally caches npm/yarn/pnpm dependencies using `actions/cache`, registers problem matchers for TypeScript and ESLint, and configures authentication for private npm registries. The architecture is a composite action with a main entry point (`src/setup-node.ts`) and a post-job cache-save entry point (`src/cache-save.ts`), both compiled via ncc. The codebase is well-structured with abstract base classes for distribution handling, a factory pattern for selecting the correct distribution, and separate modules for cache utilities and authentication. It relies on the `@actions/*` toolkit libraries and external services (GitHub node-versions manifest, nodejs.org dist mirrors).
Maple Mono is an open-source monospace programming font with an extensive build system. The repository is primarily a font engineering toolchain: it takes Glyphs source files, generates variable and static TTF/OTF/WOFF2 fonts, auto-hints them, patches in Nerd Font icons, merges Chinese/Japanese base fonts, and produces dozens of variants (hinted/unhinted, ligature/no-ligature, narrow/slim, etc.). The build is orchestrated by Python scripts (`build.py`, `task.py`) and supports customization via CLI flags or a `config.json` file. A web playground service (`in_browser.py`) allows users to build custom fonts in-browser with selected OpenType features. The codebase is not a running service but a build toolchain with a strong focus on OpenType feature engineeringβfeaturing a custom AST-based DSL for programmatically generating ligature, character variant, and stylistic set feature files. The repository is well-maintained with active GitHub Actions workflows, though the v7 branch is explicitly a frozen snapshot of the current stable release (development continues on the `variable` branch).
This repository provides a complete, production-ready solution for converting LoctekMotion/FlexiSpot standing desks into smart, Home Assistant-integrated devices. The core of the project is a set of ESPHome YAML packages and custom ESPHome components that communicate with the desk controller over a serial (UART) interface via an RJ45 connection. The firmware decodes the desk's proprietary 7-segment display protocol to report height as a sensor, and sends commands (up, down, presets, memory, alarm, child lock) to control the desk. It supports both direct connection (using a spare RJ45 port) and pass-through configurations (intercepting the keypad line). The repository includes well-structured, parameterized YAML configurations for ESP32 and ESP8266, custom C++ sensor components for height and keypad command decoding, and Python code generation stubs for ESPHome. Archived Arduino sketches and a Raspberry Pi Python script are also provided for alternative implementations. The project is actively maintained, uses modern tooling (uv, ruff, pre-commit, CI), and demonstrates strong embedded systems and protocol reverse-engineering skills.
This repository is a comprehensive configuration and extension set for Anthropic's Claude Code AI coding assistant. It enhances Claude Code with a sophisticated hook system for event-driven automation, a specialized agent delegation system for task-specific AI assistants, and an activity tracking mechanism that injects context-aware guidance based on user behavior. The system also includes templates for generating project documentation and integrates with external services like Pushover for notifications. While it contains example documentation for a game project (likely 'Mystica'), its primary purpose is to provide a customizable development environment for AI-assisted software engineering.
oidc-spa is a production-grade OpenID Connect (OIDC) client library designed for browser-first single-page applications. It implements the Authorization Code + PKCE flow with security-first defaults including in-memory token storage, DPoP (Demonstration of Proof-of-Possession) support, token substitution to prevent token leakage, and browser runtime freezing to resist monkey-patching attacks. The library provides framework-specific adapters for React, Angular, Next.js, Nuxt, TanStack Start, and React Router, each offering hooks, guards, and middleware for seamless authentication integration. It also includes a Vite plugin for build-time injection, a mock OIDC adapter for testing, and server-side utilities for access token validation (RFC 9068). The core is framework-agnostic, with a custom reactive event system (Evt) for state management and cross-tab synchronization via BroadcastChannel. The repository contains extensive example applications demonstrating login/logout, protected routes, role-based access, and auto-logout overlays across multiple frameworks.
DoWhy is a mature Python library for causal inference that implements a systematic four-step workflow: model (encode causal graph), identify (derive estimand), estimate (compute causal effect), and refute (validate robustness). The library supports two major frameworks: (1) classical potential-outcomes-based estimation with ATE/ATT/CATE via backdoor, frontdoor, instrumental variables, and mediation analysis, leveraging multiple estimators including linear regression, propensity score methods, doubly robust, instrumental variable, regression discontinuity, and integrations with EconML and CausalML; (2) Graphical Causal Model (GCM) based inference for what-if analysis, counterfactual sampling, distribution change attribution, arrow strength, intrinsic causal influence, anomaly attribution, and root cause analysis. The architecture is modular with abstract base classes for estimators, refuters, interpreters, and GCM components. The main entry point is the CausalModel class, which orchestrates identification, estimation, and refutation via a consistent API. Key dependencies include scikit-learn, statsmodels, networkx, sympy, pandas, numpy, and optional integrations with Torch, EconML, Causal-learn, and TabPFN. The library is production-ready with comprehensive test suites, CI/CD, and extensive documentation, and is actively maintained with a large contributor base.
MicroLink v2 is a production-ready Tailscale VPN client for ESP32-S3 devices that establishes secure WireGuard tunnels through Tailscale's control plane. It implements the full Tailscale protocol stack including Noise protocol for control plane encryption, HTTP/2 framing, DISCO peer discovery, STUN for NAT traversal, DERP relay fallback, and WireGuard encryption via wireguard-lwip integration. The architecture uses a multi-task FreeRTOS design with separate tasks for coordination, WireGuard management, DERP transmission, network I/O, and application logic, communicating via message queues and event groups. Key features include WiFi/cellular network failover with health monitoring, zero-copy WireGuard receive option for high-throughput scenarios, NVS-based peer caching with LRU eviction, and an HTTP configuration server for runtime management. The code demonstrates strong protocol fidelity through direct reference to Tailscale's Go source files while adapting to embedded constraints.
python-hyperscan is a CPython extension providing Python bindings for Vectorscan (a fork of Intel's Hyperscan), a high-performance multiple regex matching library. The library enables efficient pattern matching with statically linked binaries and cross-platform support. It exposes most of the Hyperscan C API through Python classes including Database for pattern compilation, Scan for text scanning, Stream for streaming processing, and Scratch for thread-safe scratch space management. The implementation features proper buffer protocol support, sophisticated GIL handling for both traditional and free-threaded Python models, and careful resource management. The build system uses scikit-build-core with CMake to statically link Hyperscan/Vectorscan along with dependencies like PCRE and Boost, producing binary wheels for multiple platforms.
ε°θ΄ (xiaobei) is an AI-powered multi-agent content marketing automation platform for small and medium businesses, built on top of the OpenClaw agent framework. It focuses on producing, publishing, and tracking content across major Chinese social media platforms (WeChat, Douyin, Xiaohongshu, Bilibili, etc.). The system uses a custom forked browser automation CLI (camoufox-cli) for platform interaction, a relay service for API signing, and SQLite databases for content and performance metrics. It includes multiple specialized agent crews (main, IT engineer, content producer, sales-cs) that collaborate through a modular skill system with thin wrapper scripts for agent invocation. The codebase demonstrates solid architecture with consistent patterns (DNA style analysis, workflow steps, patch application), but is heavily coupled to specific platform APIs and DOM structures, making it fragile to upstream changes. It is an actively developed, production-grade application aimed at automating social media operations.
CyberClaw is an enterprise-grade transparent AI agent architecture that implements a dual-watermark memory system, full behavior auditing, zero-trust execution via two-phase skill invocation, and a heartbeat task engine for scheduled operations. The system provides an interactive conversational interface where users can query time, perform calculations, manage files, execute shell commands within a secure sandbox, and utilize pluggable skills. All agent actions are transparently logged to JSONL files for monitoring and compliance. The architecture features clean separation of concerns with distinct modules for agent logic, skill loading, context management, logging, heartbeat tasks, and tool execution.