ContextWell
The collaboration layer for AI-native teams

Where your team, your AI, and your knowledge work together.

Claude, Cursor, ChatGPT, Codex all read and write markdown. Your team writes in Google Docs, Notion, Confluence. ContextWell is the in-between: a real-time editor on top of your knowledge base, with an MCP that plugs into every modern AI tool — so the same document is the source of truth for people and agents.

The middle layer AI's been missing

AI works in markdown.
Your team doesn't.

Your team's docs, decisions, and conversations live in Google Docs, Notion, Confluence. AI tools read and write markdown. So every AI-assisted task becomes an unnecessary and costly round-trip.

what this changes

Real-time collaboration between people and AI. On the format AI was built for.

01

A real editor for the format AI already speaks

Real-time co-editing, comments, suggestions, threads — on plain markdown. The thing Google Docs gives you, but for the format Claude and Cursor were built around. No more pasting between tools.

  • Multiplayer markdown editor with comments & presence
  • Folder structure, search, backlinks, graph view
  • The same doc your designer edits is the doc your agent writes into
02

Plug into every AI tool you already use

One MCP server exposes your workspace to Claude, ChatGPT, Cursor, Codex, Claude Code, and any custom agent. Your team picks the harness; the knowledge stays in one place.

  • Native MCP — works with the connector list you already have
  • Switch harnesses, switch models, keep your context
  • Your skills travel with you — same surface, every tool
03

Knowledge that compounds

Your team's docs, decisions, meetings, and skills become a structured workspace agents can actually reason over. Folder hierarchy, full-text search, backlinks, and a live graph of how everything connects — so context is never more than a query away.

  • Folders, search, backlinks, graph overview
  • Skills as first-class artifacts — versioned, attributed, reusable
  • Available to every harness through the same MCP
your data, portable by default

Your knowledge. Yours to take with you.

Everything you write in ContextWell is plain markdown. Export the whole workspace as a folder of .md files at any time. Sync to your local disk, push to a Git repo, mirror to Obsidian — your call.

We're a workspace, not a vault. The lock-in story we want is: you stay because the collaboration is good, the MCP is fast, and your skills work everywhere — not because we're holding your files hostage.

  • Markdown all the way down. What you see in the editor is what you'd export — no proprietary format, no lossy conversion.
  • Yours to take with you. Export the whole workspace as a folder of .md files at any time. Sync to disk, push to a Git repo, mirror to Obsidian — your call.
  • Made for humans. Real-time editing, comments, suggestions, presence — on top of plain markdown.
  • Made for AI. The same files are read directly by Claude, Cursor, Codex via MCP — no scraping, no exports, no lossy round-trips.

From producing content to delivering outcomes.

A shared workspace for your team — one format, one continuous loop between people and AI.