Answers
What are the alternatives to CLAUDE.md for sharing AI context across a team?
Three of the answers are still files. AGENTS.md is a single file that several coding agents read; Cursor's documentation says it supports it in the project root and subdirectories, GitHub Copilot reads it too, and Claude Code reads it only when the project has no CLAUDE.md. Tool-specific rule files — Cursor's .cursor/rules, Copilot's .github/copilot-instructions.md, Gemini CLI's GEMINI.md — give each tool its own instructions. An import or a symlink makes several of those files one. All three change which file a tool reads, not what the file is: text the model loads every session and an assistant can add to, with nothing saying who approved a line or whether it still stands, out of sight of anyone who does not open the repository. The alternative that changes that is to take what holds out of the file and put it where someone sees it and approves it. Arroway is that place: rules and decisions per project, each with who approved it and when it ends, read before acting by every connected tool. The install page is the first step.
Last updated October 6, 2026
The file options and what each one changes
AGENTS.md: one shared file. Cursor supports it in the project root and in subdirectories, with the more specific file applying where it sits; Copilot reads the nearest AGENTS.md; Claude Code reads it only when no CLAUDE.md is found. Tool-specific files: .cursor/rules holds .mdc files with frontmatter, Copilot reads .github/copilot-instructions.md and path-specific .instructions.md files, and Gemini CLI reads GEMINI.md by default and can be told to read other names through its context.fileName setting. Imports and symlinks: CLAUDE.md can import another file, and a symlink gives one file two names. All three answer which file a tool reads. None of them changes what a file is.
Where each of them stops
A file is read by the agents working in that repository. The person setting prices in a chat window never loads it, and neither does an agent in another repository. A line in it cannot say it is no longer true, so last month's rule reads like today's. A pull request tells you who changed the line and not whether the line still stands. And nothing in a file separates a rule a person approved from one an assistant added in a session. These limits are not a flaw in any one tool; they follow from the rule living inside a repository.
When to use Arroway
Use Arroway when the team needs to see what its assistants work from. Each rule and decision is recorded with who approved it and the condition that ends it, in one place the team can read, and every connected assistant — Claude, ChatGPT, Codex, Cursor or anything over MCP — reads what is in force before it acts, in the repository and outside it: a chat assistant, an agent in another repository, a colleague who does not use Git. What holds, about the code or about the business, goes there and not into an instruction file, where an assistant can add a line nobody approved. The first step is the install page.
What it looks like in practice
A team of five uses Claude Code and Cursor on the same repository. They keep one AGENTS.md, and CLAUDE.md imports it, so both tools read the same file. In October the product lead decides that the free plan drops from five projects to three. The engineers change the limit in code; nobody edits the file. The support assistant, running in a chat window and never in the repository, keeps telling customers five. When someone finally opens the file, it has lines nobody remembers approving: a test command that no longer exists and a note an assistant added in August about the five-project plan. In Arroway the decision is recorded once, with the product lead's name, as replacing the five-project rule, and what still holds from the file goes in the same way, approved by a person; the file is left with nothing in it. The coding agents, the support assistant and a new hire's assistant all open their next session with the three-project rule, and the history shows the old limit and when it changed.
Questions people ask about this
- Is AGENTS.md a replacement for CLAUDE.md?
- Claude Code reads AGENTS.md only when there is no CLAUDE.md, and a CLAUDE.md that imports AGENTS.md makes it read it too; Cursor and Copilot read AGENTS.md. Swapping one file for the other changes which tool reads the text, not who sees it or approves it.
- Do I still need the files if I use Arroway?
- No. What holds goes to Arroway, where someone sees it and approves it, and every connected tool reads it before acting. An instruction file with rules in it is a second source the team does not see: the model loads it every session, and an assistant can add a line to it that nobody approved.
- How do I try Arroway?
- Install it in the tool you already use — the install page has one route for Claude, ChatGPT, Codex, Cursor or any MCP tool — and sign in when the tool asks. From then on, sessions open by reading the project and close by recording what was decided.
Where this is verifiable
Cursor documentation on rules (cursor.com/docs/context/rules), GitHub Docs on repository custom instructions for Copilot (docs.github.com/en/copilot), and the Gemini CLI documentation on GEMINI.md and the context.fileName setting (geminicli.com/docs/cli/gemini-md), read on 6 October 2026; the Claude Code memory documentation and changelog, read on 29 September 2026. Each product is described only as its documentation describes it on those dates, and can change in any release. Product documentation on this site (How it works, Install), the answers on whether Claude Code reads AGENTS.md and on Arroway versus a context file, and the sanction, supersession and expiry-condition rules in the sanctioned product spec. Everything described here about Arroway is behaviour the tools apply today, not roadmap. The team in the example is illustrative.
https://www.arroway.app/en/answers/alternatives-to-claude-md-for-a-team