Coding agents are quietly locking companies into per-vendor config files — and the most prominent CEO in the AI-tooling conversation is pushing back.

Shopify CEO Tobi Lütke says he's considering banning Claude Code from the company unless Anthropic starts natively reading the shared AGENTS.md standard — a threat that has turned a one-line file format into an open fight over whose spec wins in the coding-agent market. In a post on X, Lütke said he would pull the tool unless Anthropic supports both AGENTS.md and .agents/skills, the same files already read by OpenAI's Codex, Cursor, Gemini CLI, GitHub Copilot, Devin, and VS Code.

The core complaint is what Lütke calls "split brain." Engineers at one company now run several AI coding tools side by side, and each only reads its own config folder. One agent knows how the tests run and which directories it must not touch; another, working on the same repo, has none of that context. "Today it's CLAUDE.md, tomorrow some other vendor's file — the more AI tools you add, the more fragmented the rules become." That fragmentation is why Shopify, a company that has baked AI fluency into performance reviews and now requires teams to justify any headcount AI can't cover, treats config portability as a hill worth fighting on.

Anthropic's Claude Code team member Thariq Shihipar hit back within hours, saying the company is making the tool "more modifiable" and easing access to AGENTS.md or other system-prompt settings. But he defended Claude Code's own CLAUDE.md / .claude/skills / .claude/rules structure, arguing that system prompts materially change how a model behaves and that different model families can't simply be swapped. For now, users can force Claude Code to load the shared file by adding a line to CLAUDE.md — a workaround that does little to resolve whose format becomes the default.

The tension is sharper than a mere feature request. AGENTS.md now lives in more than 60,000 open-source repos, and OpenAI donated it to the Linux Foundation's Agentic AI Foundation in December 2025 — a body Anthropic itself co-founded, having contributed the MCP protocol. The awkward part is that Anthropic helped stand up the open agent infrastructure while its marquee coding tool still refuses the shared file natively. The New Stack reported Anthropic had already closed the developer feature request for AGENTS.md support, making Lütke's ultimatum all the more pointed.

This is the quiet governance war of the agent era: the models get swapped all the time, but the rules, permissions, and workflows a company writes down are the actual moat — and whoever owns the file format controls it. Anthropic betting that a superior model beats an open standard is the central wager, and Shopify is the kind of flagship customer that tests it.

What to watch next: whether Anthropic folds native AGENTS.md support into the next Claude Code release, or holds its line and risks more big-ticket defections.

Is a single shared config standard good for the coding-agent market, or does it just hand the advantage to the biggest vendor? Tell us in the comments.

Sources: AGENTS.md feature request (GitHub)

Akhetonics says its all-optical CPU reaches a customer in 2026

Light-based computing keeps promising more than it delivers — but one Munich startup has just put a date on its bet, and the interview laying it out is doing the rounds on Hacker News this week. Akhetonics says it will deploy its first commercial machine with a major customer by the end of 2026, with several more planned for 2027. The company, founded by Michael Kissner and Leonardo Del Bino, is building a computer where data enters as light, is switched as light, and circulates through memory

The Week in AI — October 5–11, 2026

Every big claim this week turned out to rest on fine print more interesting than the headline: revenue only the company reporting it can define, safety tests sandboxed while the product keeps the web, and a Pentagon phase-out nobody would confirm until reporters kept asking. The week's top 5 1. OpenAI's revenue was $20 billion below the numbers everyone quoted — and the gap was definitional. The Financial Times reported Thursday that OpenAI's annualized revenue runs roughly $20 billion unde

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Russia's largest tech company is learning what the AI era's infrastructure war looks like from the receiving end — three data centers in four days, and with them much of the cloud layer Russian businesses run on. A Ukrainian drone strike knocked out Yandex's data center in Vladimir early Sunday morning, the third of the company's facilities hit since October 8. The site — reported at roughly 50 MW and designed for about 2,880 server racks — stopped operating completely after the attack, Yandex

Agent teams cost up to 5x more, barely score higher

The multi-agent hype train hit a benchmark this weekend — and the grid and the trucking regulators had quiet weeks of their own. Vals AI put agent teams head-to-head with single agents on its Vibe Code Bench, and the teams cost between 1.8 and 5.1 times more for almost no extra quality. The evals company ran GPT-6 Sol and Claude Opus 5.5 solo and in teams across 50 apps at two reasoning efforts; out of four comparisons, only one was statistically significant — Sol at medium effort, where the t

Read more

Akhetonics says its all-optical CPU reaches a customer in 2026

The Week in AI — October 5–11, 2026

Drone strike shuts a third Yandex data center, taking YandexGPT offline

Agent teams cost up to 5x more, barely score higher