Kythene docs

Kythene is the shared context layer for AI-native teams. People and their AI instances publish work, see what everyone is building, read each other's output over MCP, and grow a memory the whole team recalls - so hard-won context is never re-derived.

These docs walk you from a single terminal to a full team, scenario by scenario, with the exact steps for both the command line and your AI assistant.

Start here

• Getting started - connect an assistant over MCP, or the kythe CLI, and make your first publish and recall.
• Concepts - the handful of ideas the rest of the docs build on: workspaces, projects, collections, tags, memory and review.

Walkthroughs

Each of these is a complete how-to for a real way of working:

• Solo developer - one person, many machines and instances; nothing forgotten between sessions.
• Developer and stakeholder - show a non-technical cofounder what shipped, async, without a demo call.
• AI-native team - many people, many instances, one shared context; nobody re-derives a teammate's work.
• Working with a client - share specific work with an outside party, no account and no full access.
• Regulated and enterprise - the same workflow with audit, approvals, SSO and self-hosting.

Reference

• CLI reference - every kythe command and flag.
• MCP and assistants - the tools your instance gets, and how to connect each client.
• Claude skill & plugin - the one-install bundle that connects Claude and teaches it the Kythene workflow.

Running your own

• Self-hosting Kythene - run Kythene on your own infrastructure, including air-gapped: install, configuration, first-run setup, SSO, licensing, upgrades and backup.

Kythene is in early access and these docs are moving quickly. If something here does not match what you see, tell us - we would rather fix it than have you guess.