drawio-skill is a comprehensive AI agent skill for generating, editing, synchronizing, inspecting, testing, and publishing editable draw.io diagrams from natural language, codebases, infrastructure-as-code, and data schemas. It provides a unified CLI (`diagramctl.py`) orchestrating a pipeline of 39+ specialized scripts covering importers (Python, JS/TS, Go, Rust, Terraform, K8s, Docker, SQL DDL, OpenAPI, CI pipelines), auto-layout (Graphviz, ELK, deterministic sequence/C4/tube-map), shape search (10k+ official icons), AI/LLM brand logos (321 brands), style presets, diagram diff, time-lapse, heat maps, and export (PNG, SVG, PPTX, Mermaid, HTML viewer). The architecture is centered on a versioned Diagram IR (JSON Schema) that separates meaning, provenance, and geometry, enabling incremental sync preserving manual layout, multi-view projection, architecture contract testing (Diagram-as-Test), what-if failure simulation, and accessible Story Mode. It integrates with agent platforms via a SKILL.md and an optional MCP server (`diagramctl_mcp.py`) exposing nine tools over stdio, and supports CI integration through GitHub Actions for PR diff and architecture test gates. The project is actively maintained with semantic versioning, 174 unit tests, and a changelog tracking releases through v3.2.0.
Petastorm is a library for loading Apache Parquet datasets into machine learning training frameworks, primarily TensorFlow and PyTorch. It provides a reader API that supports parallel I/O, row-group shuffling, predicate filtering, and on-the-fly transformations. The library also includes a Spark Dataset Converter to materialize Spark DataFrames as cached Parquet for training. Architecture: modular with worker pools (process/thread), filesystem abstraction (local, HDFS, S3, GCS), schema system (Unischema with codecs), and shuffling buffers. It features CLI tools for benchmarking, metadata generation, and dataset copying. The codebase demonstrates strong engineering practices with clear separation of concerns, but shows signs of aging: reliance on TensorFlow 1.x, Python 2 compatibility, and pinned dependencies that may cause drift. The project has a large test suite with legacy dataset fixtures for backward compatibility, but maintenance has slowed significantly with the last release (v0.13.1) and minimal recent commits.