About Rness
Who makes Rness, what it is and what it is not, and where its code lives.
Rness is one source of truth for every AI coding agent of a GitHub organization. Your organization's standards, decisions, specifications and plans live as Markdown in one repository, .rness, and the rness command writes the part that applies into the AGENTS.md of every repository. Claude Code, Codex, Cursor and GitHub Copilot already read that file, so nothing changes in how a team works.
Who makes it
Rness is made in Paris, France, by Cédric Magne (LinkedIn), with a small team. It is open source under the MIT licence: the code, the issues and the discussions are public at github.com/rness-dev, and the packages @rness/cli, create-rness and @rness/create are published on npm from that repository.
Rness is built with Rness: its own decisions, specifications and plans are kept in a .rness workspace and move on a GitHub Project as the work happens. This site describes what the published command does today; whatever it shows that is not built yet carries the words "not shipped yet".
What it is, and what it is not
- A command-line tool on npm,
rness, for Node 22.17 or later, with git. - A local MCP server,
rness mcp, that serves a workspace's context to any MCP client, read-only, started in the workspace by the agent. - A Claude Code plugin with four commands,
/rness:status,/rness:adr,/rness:specand/rness:plan, written into each repository byrness sync. - Not an agent: Rness never runs a model, and sends none of your code or context anywhere. A workspace is files in your repositories, on your machine and on your Git provider.