Tag
The article discusses the value of a hand-picked repository of AI agent skills, contrasting it with AI-generated content, and highlights the challenges of finding and maintaining useful skills across various companies' separate repositories.
The author has developed an agent pipeline to sync the bookmark contents from the Chrome browser to a personal website for better management and sharing of interesting resources.
Art museums are increasingly using data and analytics to inform curation and improve visitor experiences, moving from ivory towers to participatory centers.
A web tool that curates over 3,100 historically popular Hacker News links, categorized by a hall of fame induction system based on submission frequency and breakout events.
A personal essay reflecting on the value of curated, high-quality non-fiction books as a counterpoint to low-quality AI-generated content, lamenting the decline of traditional library browsing and the rise of AI slop.
SkillCorpus presents a framework to consolidate, curate, and evaluate the open skill ecosystem for LLM agents, demonstrating consistent performance gains across multiple benchmarks through retrieval-augmented skill integration.
A new skill/plugin that uses X MCP to let AI agents track high-signal X accounts and generate a local HTML feed of top AI news, papers, and projects.
A researcher shares a daily automation that curates AI news from high-signal X accounts into an HTML artifact using X MCP tools and research agents.
A website that presents 30 essential machine learning papers curated by Ilya Sutskever in a beginner-friendly format.
The author advocates for LLM Wikis as a valuable application of AI, showcasing their PaperWiki project that uses agents to automatically curate and maintain a knowledge base of research papers, improving signal-to-noise ratio and enabling cutting-edge research.
REAP is an automated pipeline that curates production-derived benchmarks for coding agents from real developer-agent sessions, using LLM-based classification and stability checks to ensure reliable evaluation without manual labeling.
Kira Community is a tool for curating moments with hashtags, as featured on Product Hunt for art and social media integration.
Introduces the anbeime/skill project, which automatically collects and categorizes Claude Skills from GitHub, offering 243 official and local skill packages, with automatic updates every 24 hours.
This paper argues that for large enough models, unfiltered data can improve generalization by providing weak perturbations, contrary to the common assumption that only high-quality filtered data is beneficial. The authors caution that harmful conditional shifts can still damage models, but over-curation may remove useful perturbations.
A curated list of 11 emerging AI tools from countries outside the US and China, expected to gain traction in 2026. The tools are described as lighter, more niche, but powerful.
AgenticBrew is a new AI news aggregator that pulls from hundreds of sources and clusters related stories. The creator is now exploring a personalized learning path feature to help users improve their AI skills based on role and literacy level, and is seeking community input.
Weekly summary of notable AI research papers from May 31 to June 7, including LEAP, AutoLab, and scaling laws for agent harnesses.
BestBlogs has compiled 375 WeChat Official Account RSS feeds that are still being updated, covering multiple fields such as artificial intelligence, software programming, and business technology. It also provides an OPML file for import into RSS readers, aiming to help users independently select information sources and combine them with AI for filtering.
This paper introduces a deliberative curation protocol for multi-agent knowledge bases, addressing governance gaps such as agent statelessness and sycophancy. It evaluates the protocol via simulation, showing improved resilience under adversarial conditions.
This paper introduces BioConCal, a supervised scorer that uses inference-time panel and candidate features to rank biomedical entity candidates surfaced by LLM panels, significantly improving over raw agreement for curator triage.