Skip to main content
A project is one archive of coding-agent history, and it is tied to a directory on disk. resumecontext init does two things that have to happen together: it creates the project in the cloud, and it writes a marker file in the directory recording which project this is. Everything afterwards, including sync, sharing and the MCP URL, reads that marker to know which project it is talking about.
This is why projects cannot be created from the website. A browser has no directory to mark, so a project made there would have nothing syncing into it.

​
One directory, one project

init marks the exact directory you run it in, and never walks up to an enclosing git root. That is deliberate: it lets a subfolder of a monorepo be its own project with its own archive and its own access list.

​
A cloned repo that is already a project

If the marker file is committed to git, a teammate clones it, accepts their invite (in the CLI or dashboard), and runs resumecontext sync once from that clone. The first sync registers the directory with their local daemon. From then on their local history syncs automatically into the shared project. They should not run init: a directory without the shared marker would create a separate project instead of joining yours.

​
Renaming

A project is named after the directory you ran init in. The owner can rename it from the dashboard: Rename on the project’s card. Only the name changes; the project id, so every member’s marker file and MCP URL, stays the same. Coding agents pick up the new name the next time they connect.

​
Deleting

Deletes the project and all the conversations synced into it, for everyone. Owner only, and it cannot be undone. You can also delete from the dashboard.