![3Blue1Brown: The last IMO problem AI could not solve [video]](https://hntoday.pixelshot.io/screenshots/20260918/1b1964c6ce824fe0_tn.jpeg)
3Blue1Brown: The last IMO problem AI could not solve [video]
The video discusses a challenging IMO problem that AI was unable to solve. It highlights the limitations of current AI problem-solving capabilities. (youtube.com)
Anthropic finally adds AGENTS.md support to Claude Code
twitter.com41 points7 comments
Anthropic has added explicit AGENTS.md support to Claude Code in version 2.1.277: if a folder lacks a CLAUDE.md file, Claude will now look for and use AGENTS.md as the project instruction file. That fallback behavior is enabled by default but can be toggled in /config, so teams can opt in or out per environment. The change is a narrow but practical update that affects how Claude discovers and applies project-level agent instructions during code-focused sessions.
The AGENTS.md capability is implemented as a built-in Claude Code mod and is part of a broader Claude Code mods system that will let developers create custom project instruction behaviors. Anthropic published the mod’s source on GitHub and pitched this as the first of many customizable harnesses for managing agent behavior across projects. The announcement drew mixed reactions from the community, including some criticism that this is basic or overdue functionality, but it concretely establishes a standard file name and an extensibility path for team-specific agent configurations.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
More in AI
![Speaking with the Mind – Neuralink [video]](https://hntoday.pixelshot.io/screenshots/20260918/9f12fd89b0bb495d_tn.jpeg)
Speaking with the Mind – Neuralink [video]
Neuralink demonstrates a brain-machine interface that allows direct communication between the human brain and computers. The technology aims to enable new ways of interacting with digital devices and treating neurological conditions. (youtube.com)
Cache-to-Cache: Direct Semantic Communication Between Large Language Models
Cache-to-Cache (C2C) enables direct semantic communication between large language models by projecting and fusing their KV-caches, avoiding text-based exchange. Experiments show C2C improves accuracy by up to 14.2% and reduces latency by 2.5 times compared to traditional methods. (arxiv.org)
Gov. Newsom signs executive order targeting AI safety 'before it's too late'
Governor Gavin Newsom has signed an executive order aimed at ensuring AI safety. The order emphasizes proactive regulation to address potential risks associated with artificial intelligence. (ktla.com)
Who's a Better Writer: A.I. Or Humans? NYTimes Quiz
The NYTimes quiz compares the writing abilities of AI and humans to determine which is better. Results show a majority favoring AI, despite perceptions of AI writing being inferior. (nytimes.com)
Europe's AI firms, playing catch-up, challenge US calls for slowdown
European AI companies are working to catch up with their US counterparts amid calls for a slowdown in AI development. They are challenging the US push for regulating or delaying AI advancements to ensure competitiveness. (reuters.com)