Setup
Collect AI-attribution on every commit, Pull Request, and line of production code.
To collect telemetry and metrics from across your team you need to install the CLI on every developer endpoint and connect your team's codebases.
- Connect source control — read authorship data across your org and track AI-authored code by pull request, repository, and team.
- Install Git AI on developer machines — collect coding agent telemetry (prompts, session traces, AI authorship) from every agent session.
Connect Source Control
Developers with the Git AI CLI installed attach an authorship log to every commit via git notes. Connecting a source control provider lets Git AI read those notes at the organization level and track AI-authored code in aggregate — by pull request, repository, and team.
Supported Providers
| Provider | Status |
|---|---|
| GitHub | ✅ Supported - cloud and self-hosted |
| GitLab | ✅ Supported - cloud and self-hosted |
| Bitbucket | ✅ Supported Cloud. ❌ Data center not supported |
| Azure DevOps | ✅ Supported |
Connect an SCM Organization
Prerequisites:
- Admin access to the SCM organization or group. Non-admins who attempt to connect will trigger an approval request sent to an org admin.
- An active Git AI for Teams account — start your trial here
Steps:
- Open Settings > Connect SCM in the Git AI dashboard
- Select the SCM provider and authenticate
- Choose one or more organizations or groups to connect
Once connected, Git AI begins processing authorship data for every repository in the organization. No per-repo configuration is required.
Permissions and Security
What Git AI accesses
| Permission | Purpose |
|---|---|
| Membership read | Link commit activity to members of your organization |
| Pull request read | Checkout, pull git notes, and run git ai blame |
| Push (for notes) | Push new git notes for squashed, rebased, and merged commits |
| Webhook events | Receive webhooks for pull request status changes |
Data handling and transport security
- Clones run on isolated, ephemeral compute. Repository data is deleted immediately after the operation completes. Git AI does not store source code.
- Push access is scoped exclusively to
refs/notes/ai— the Git AI notes namespace. Git AI never pushes to branches and cannot bypass branch protections. - Incoming webhooks are validated using provider-signed secrets. Secrets are rotated following each SCM provider's best practices.
For organizations that require full control over infrastructure and data, Git AI supports self-hosting.
Install Git AI on Developer Machines
Coding agent telemetry — prompts, session traces, AI authorship data — exists on developer machines. Installing Git AI on every developer laptop enables collection of this telemetry from every agent session automatically.
There are two ways to distribute Git AI to developers:
- Share an install link — best for PoCs and small teams
- Distribute via MDM — best for enterprise-wide rollouts
For a proof-of-concept evaluation, start with an install link for the first 10–20 engineers. Switch to MDM distribution when rolling out to the full organization.
Share an Install Link
Best for PoCs and small teams.
In the Git AI dashboard, open the Invite your Team tab and copy the invite link. Share it in a document, email, Slack message, or any other channel.
When a developer opens the link, they authenticate with their SCM and get a personal install command to run on their machine. Once complete, telemetry from every supported coding agent begins flowing to the dashboard.
Distribute via MDM
Best for enterprise rollouts.
For large-scale deployments, push Git AI across the organization with an MDM solution — no manual install per developer. See the MDM deployment guide for requirements, install steps, and configuration.