
I kept running into the same problem: my team's AI context lived in plans.md / claude.md / spec files, but there was no good way to co-edit them, and agents only ever saw old pasted snapshots. So I fixed it building easymd, completely free
So I built easymd: real-time collaborative markdown editing, like Google Docs, except the document is an actual .md file and an AI agent can edit it live alongside you.
- Real-time multiplayer editor (live cursors, CRDT sync, no merge conflicts)
- The file on disk stays canonical — edits sync straight back, no drift
- MCP server built in: connect an agent in one line; it can read, create, and update the same docs you're editing. Its changes appear live in your editor, and your edits are instantly visible to it.
- Clean markdown also saves a ton of tokens vs feeding models HTML/DOCX/PDF
- One-line CLI to start: npx easymd open CLAUDE.md
The idea is one shared, always-current file that both humans and agents work from, instead of copy-pasting context around.
Would love feedback! Especially on the agent-as-collaborator workflow and what'd make it actually useful in your setup.