AI Pulse · AI Trends Pulse

Claude Code now reads an AGENTS.md file when a project doesn’t have a Claude.md file. That sounds minor, but it points to a useful shift in how companies set rules for AI coding agents. Instead of writing separate instruction files for each tool, teams can increasingly keep one shared set of project guidance that different agents can understand. Anthropic documented the change in its Claude Code changelog.

An AGENTS.md file can hold the practical context an AI agent needs before it touches a codebase. Think coding conventions, test commands, repository structure, deployment rules, and things the agent must not change. If your developers use more than one AI coding tool, a portable instruction file reduces the chance that each tool works from a different version of the truth. The update drew 709 points and 262 comments on Hacker News, which suggests developers are paying close attention to cross-tool standards.

For owners, the bigger point is governance. AI agents become more useful when they operate within clear boundaries, and less useful when every person has their own undocumented setup. If your team is already using Claude Code or considering it, ask whether project-level AI instructions are written down, reviewed, and stored with the code. This is the kind of thing we build into an AI command centre, giving teams a clearer way to manage how AI tools work across the business.