Hacker NewsTuesday, April 15

Harvard rejects federal government demands regarding campus antisemitism policies, viewing them as overreach that threatens university independence and First Amendment rights. While committed to fighting antisemitism, Harvard asserts its autonomy in managing campus affairs and maintaining academic freedom.

This guide provides comprehensive information about equity compensation, including how companies grant partial ownership to employees in exchange for work. It covers complex topics like restricted stock, stock options, and tax implications, aimed at helping employees, hiring managers, and founders make informed decisions about equity compensation in U.S. corporations.

The article discusses the Model Context Protocol (MCP), highlighting its role as a standard for connecting third-party tools with AI assistants while examining various security concerns, limitations, and vulnerabilities in its implementation. Key issues include authentication challenges, potential exploits, and the need for better user protection.

A developer shares their experience building a simple but effective AI butler named Stevens that sends daily briefings via Telegram, including calendar schedules, weather forecasts, and reminders. The system uses basic SQLite storage and scheduled tasks to manage memories and deliver personalized updates in a formal, butler-like tone.

OpenAI announces GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano for their API platform, featuring major improvements in coding, instruction following, and context handling capabilities. The models outperform previous versions with enhanced performance across benchmarks while offering lower costs and better real-world utility.

DeepSeek announces plans to collaborate with existing open-source projects after deciding against open-sourcing their full internal inference engine due to codebase complexity and resource limitations. The company will focus on contributing modular features and optimizations while maintaining support for future model releases.

Whenever is a Python library designed to handle datetime operations more safely and efficiently than standard alternatives. It offers DST-safe arithmetic, type-safe API, and improved performance while preventing common datetime-related bugs. The library provides explicit types for different use cases and is available in both Rust and pure Python versions.

Fennel is a programming language built on the Lua runtime, designed to address Lua's shortcomings while maintaining its benefits. It features a lisp-like syntax, stricter variable handling, improved table notation, and additional features like destructuring and pattern matching, while preserving Lua's simplicity and speed.

A financial reporting tool that automatically generates earnings summaries for over 1000 US companies immediately after their SEC filings. The service tracks major companies like Apple, NVIDIA, and JPMorgan Chase, providing quick access to earnings reports and financial performance analysis through their website signalbloom.ai.

A woodworker describes attending the 39th Kezurou-kai event in Japan, where participants compete to create the thinnest wood shavings using Japanese planes. The two-day event featured preliminary rounds with hinoki wood and a final competition using sugi wood, where competitors demonstrated exceptional skill in achieving incredibly thin, consistent shavings.