AI sessions
Turn on team knowledge sync
Updated July 25, 2026
Team knowledge sync is what makes your agent start a session already knowing your team. Without it, your agent has Milestone's tools but no reason to reach for them. With it, your team's context arrives on your first prompt.
It's off until you turn it on, deliberately, on each machine.
What it does
Two things, in your Milestone repositories only:
- On your first prompt, your team's ranked context is fetched and added to the session - conventions, decisions, and the gotchas behind them.
- When a session ends, the knowledge it actually used is reported back, so the things that keep proving useful rank higher next time.
That second half is why recall sharpens over time rather than staying flat.
Turn it on
Open your connection settings
Go to Settings → Connections and scroll to Team knowledge sync.
Enable on this machine
Turn on Enable on this machine. Milestone explains exactly which file it edits before changing anything.
Allow the change
Click Allow & enable. One click turns it back off later.
Connect Claude Code first
The switch stays disabled until your Claude Code connection reads Healthy - knowledge sync uses that same credential. See Connect Claude Code.
Using Codex too?
Codex has its own switch, Team knowledge sync for Codex, just below. It works the same way and reuses the same credential.
Check it's working
Open a terminal session inside a bound repository and ask your agent something your team has decided but a stranger wouldn't know. If it answers from your workspace without being asked to look it up, sync is live.
It only acts on your Milestone repos
Sync checks the repository's git remote against the repos in your workspace. In any other folder it does nothing at all - personal projects and unrelated code are never touched.
Related
- Bind a repo - sync only acts on bound repos.
- Troubleshoot agent connections - if it's on but quiet.
Ready to try Milestone?
Give your team and its AI agents one shared memory.