Use case

A decision made once, known by everyone's agents.

You primed your Cursor install with the team's architectural decisions. Priya didn't. Now her model is arguing for Redis on the job queue — a question you settled three months ago in a Slack thread she never saw. Every teammate's agent is re-learning the same context, and disagreeing about it.

CTXone Team puts one shared Hub at the center of your toolchain. Everyone's AI tools read and write the same memory graph, so context is shared by default — not copied per laptop, not re-primed per person, not trapped in one engineer's chat history.

How it works

# One person stands up the shared hub (or run it as a service)
ctx serve --http --auth-token $CTX_TOKEN

# Everyone else points their tools at it
ctx init --transport http --mcp-url https://ctx.internal:7000/mcp \
    --auth-token-env CTX_TOKEN

From that point on, ctx init wires each person's Claude Code, Cursor, VS Code, and the rest into the same hub. What one engineer teaches CTXone, the whole team's agents already know on the next turn.

What changes

  • No more divergent context. A settled decision lives in one graph everyone reads. Nobody's model re-litigates a choice the team already made.
  • Provenance points to people. On a team hub, ctx blame and ctx why-did-we attribute each fact to the real user who wrote it — not an anonymous agent.
  • Branch and merge context like code. Sandbox a speculative direction on a branch, review the diff, and merge it into the shared graph when the team agrees.
  • Plans everyone can see. Work-in-flight lives on the hub, so "what's next" is a shared answer, not a private checklist in someone's session.

Who it's for

Teams already using AI coding tools daily who are tired of every member's agent starting cold. If you've ever watched a teammate's model undo a decision you made for good reasons, this is the fix.

Next: CTXone Team overview  ·  How branches work  ·  All use cases