Cog is a native macOS audio player with an exceptionally broad format support, uniquely specializing in video-game music emulation and obscure audio codec playback. Its architecture is built on a clean, modular plugin system (CogDecoder, CogSource, CogContainer protocols) that integrates dozens of vendored emulator cores (vgmstream, GME, OpenMPT, lazyusf) and general decoders (FFmpeg, TagLib). The app features a composable DSP chain for real-time effects (equalizer, HRTF binaural, tempo/pitch shifting), high-performance threading primitives (CogSemaphore), and a modern Model Context Protocol (MCP) server for remote control by AI agents. This is a classic case of a well-crafted application core coupled with a massive, critical dependency on vendored third-party code; the plugin protocols, DSP nodes, synchronization primitives, and MCP bridge are high-quality, reusable components, while the collection of emulator cores introduces significant licensing and maintenance complexity.
This repository curates and distributes Maxar Open Data satellite imagery for disaster response events. It provides a structured catalog of pre- and post-event high-resolution imagery (WorldView, GeoEye) in multiple formats: GeoJSON footprints, union geometries, MosaicJSON tile indices, and TSV metadata tables. The core pipeline (`maxar_data_catalog.py`) fetches the official Maxar STAC catalog, iterates over event collections, and generates these derived files, which are then committed to the repository. A Streamlit web app (`streamlit_app.py`) offers interactive browsing of footprints and mosaics. The repository is primarily a data asset with a supporting ETL pipeline and a thin frontend, making it a hybrid of data curation and a lightweight application. The data is organized by disaster event (e.g., hurricanes, earthquakes, floods) and includes rich acquisition metadata (platform, GSD, cloud cover, viewing angles) suitable for geospatial analysis and machine learning.