Shopify CEO Threatens Claude Code Ban Over AGENTS.md Support Dispute

Shopify CEO Tobi Lütke recently took to X (formerly Twitter) to voice a strong stance: he is considering banning the use of Claude Code within the company unless the tool adds native support for AGENTS.md and .agents/skills. This move stems from a growing "split brain" problem at Shopify, where engineers using multiple AI programming tools—such as Claude Code, Codex, and Cursor—face inconsistencies because each tool relies on its own configuration files, leading to conflicting rule adherence across the same project.

In response, Claude Code team member Thariq Shihipar acknowledged the issue and stated that Anthropic (the company behind Claude) is working to make Claude Code more adaptable, allowing easier integration with AGENTS.md and custom system prompt settings. However, he explained that Claude Code’s proprietary system—using CLAUDE.md for project rules, .claude/skills for skills, and .claude/rules for detailed guidelines—is rooted in the need for model-specific prompt formats, as different AI model families perform best with tailored prompts.

A temporary workaround exists: adding a line "@AGENTS.md" in CLAUDE.md allows Claude Code to load the AGENTS.md file at startup. But the core dispute lies in who should set the default standard for configuration files in AI programming tools.

AGENTS.md acts as an onboarding manual for AI "programmers," containing critical project details like structure, installation steps, test commands, code standards, and security constraints. It has been adopted by over 60,000 open-source projects and supported by major tools including Codex, Cursor, Gemini CLI, GitHub Copilot, Devin, and VS Code. Notably, OpenAI donated AGENTS.md to the Agentic AI Foundation (under the Linux Foundation) in December 2025, with Anthropic as a co-founder of the foundation—creating an interesting contradiction as Claude Code still lacks native support for the file.

Shopify’s emphasis on this issue reflects its deep integration of AI into daily operations. The company requires all employees to be proficient in AI, includes AI usage in performance evaluations, and mandates teams to prove tasks can’t be done by AI before hiring new staff. Shopify also uses multiple AI tools and has an internal security system called Dispatch, which leverages different agents to scan for vulnerabilities and cross-verify results. For Shopify, preserving enterprise rules and processes across changing AI models is key—hence the push for a standardized configuration file.

The controversy highlights the urgent need for standardization in AI programming tools. As more enterprises adopt multiple AI assistants, consistent rule adherence becomes critical to avoid inefficiencies and errors. Claude Code’s ability to adapt to AGENTS.md could determine its future in environments like Shopify.

Sources

• ChinaVenture (https://m.pedaily.cn/news/568179)
• Xinzhiyuan WeChat Official Account