Pooled markdown sync
Multiple contributors sync local markdown folders to one central server. You write to yours, read from everyone's. Plain files on disk, indexed for hybrid search.
Install
curl -fsSL https://raw.githubusercontent.com/collaborator-ai/seedvault/main/install-cli.sh | bash
Uninstall
curl -fsSL https://raw.githubusercontent.com/collaborator-ai/seedvault/main/uninstall-cli.sh | bash
How it works
1
Write markdown. Your thinking lives in folders on your machine. Edit with anything.
2
Sync to the vault. A background daemon watches your directories. Changes sync automatically.
3
Everyone reads. All contributors get read access to the full vault. Agents included.
Core concepts
Vault
A single server deployment. One storage root, one set of API endpoints.
Contributor
Your namespace. One owner, one write token, one directory on the server.
Collection
A local folder you sync. ~/notes becomes notes/ on the server.
Token
sv_... bearer token. Write to yours, read from all.
CLI
sv init Interactive setup sv add ~/folder Add a collection to sync sv start -d Start syncing (background) sv status Daemon, collections, server sv ls List files in your contributor sv cat notes/foo.md Read a file from the vault sv collections List configured collections sv stop Stop the daemon Technical
- · TypeScript + Bun + Hono + SQLite
- · Plain markdown files on the filesystem
- · Hybrid search via QMD (BM25 + vector + LLM re-ranking)
- · SSE event stream for real-time updates
- · Self-hostable (Docker, Fly.io, anywhere)
- · Open source under MIT