Sharing skills with your team
Share a skill once and every teammate gets it — definitions sync across the team while credentials stay local. The day-to-day team workflow.
The point of a team is that good work spreads. When one person builds a skill that works, they share it once and the whole team has it.
Sharing a skill
- Open Skills and select the skill you want to share.
- Click Share in the skill's header.
That's it. The skill gets a Shared badge, and it's pushed to your team's library. Teammates receive it the next time their app syncs (on launch, or when they switch into the team). You'll see the sync happening in the status bar at the bottom of the window — "Syncing team library… → Team synced."
To stop sharing, click Shared again to toggle it back to personal. It's removed from the team library; teammates' local copies are theirs to keep or delete.
What teammates receive
A shared skill carries its definition — the name, description, selected tools, system prompt, and example prompts — plus provenance: who shared it and when.
It does not carry any credentials. If the skill uses tools that need an API key or token, each teammate supplies their own (stored in their OS keychain, never synced). This is what keeps sharing safe: the recipe travels, the keys never do.
Telling shared skills apart
In your skills list, shared skills show the Shared badge alongside the normal source tag. Open one and the detail pane shows "Shared by [name] · [when]" so you always know where a team skill came from.
Team analytics
Once you're on a team, your dashboard shows the team's aggregate analytics — tool usage, latency, and skill performance across everyone — marked with a Team badge. This is team-wide institutional knowledge: which skills actually work, where time goes, accumulated over time. Switch teams (or back to no team) from the user menu and the stats follow.
A note on privacy
Sharing a skill shares its definition with your team, nothing more. Your prompts, run outputs, and API keys are never shared. See Teams for the full shared-vs-local breakdown.