Reports

Reports are dashboards built from Git AI's AI-attribution data — adoption, token spend, and pull request outcomes — scoped to whoever you want to look at. Git AI ships several standard reports so you get answers on day one, and you can filter, save, and name your own.

Standard reports

Every organization starts with pre-built reports covering the questions teams ask first.

Org-wide report

Auto-created for every organization, this is your top-level view of AI usage across the whole company. Its panels group into a few themes:

• Adoption — AI-Generated Lines, Contributors Using AI, Sessions, AI-Assisted PRs, % AI Committed, and % AI Merged to Production.
• Code lifecycle — the Generated : Committed and Generated : Production ratios, plus a Sankey diagram tracing AI code from generation through commit to production.
• Agents & tokens — Agent Usage (by sessions and by contributors), Generated Lines by Agent, Token Usage, and Token Cost.
• Attribution — Committed Lines (AI vs Human) and a Contributor Leaderboard.

Per-contributor report

The same metrics scoped to a single person, plus a table of their recent sessions. Useful for 1:1s, onboarding checks, and understanding individual adoption.

Pull request report

A running, always-current list of AI-assisted pull requests across your organization, filterable by repo. Each row shows the PR title and state, % AI authorship, who opened it, and the primary agent and model used. Open any PR to drill into its detail — including cost per PR — so you can tie spend and AI authorship back to specific shipped work.

Live report

A real-time view of what agents across your org are doing right now. A timeline lays out active and recent sessions in lanes color-coded by agent, with a live sessions table beneath showing the contributor, repo, agent, and lines generated, committed, and merged. Use it to see AI activity as it happens.

Filtering reports

Every report can be scoped down to exactly the slice you care about. Filters apply globally to the whole report, so all panels re-scope together:

• Contributors — focus on specific people.
• Repos — one or more repositories.
• Teams — group-level breakdowns.
• Agents — Claude Code, Cursor, Copilot, and the other tools your team uses.
• Date range — narrow to any window.

Accessing the raw data

Reports are just one view on top of your data. You can also get at the underlying records directly:

• API — query the same AI-attribution data programmatically to build your own dashboards, alerts, or integrations.
• Direct SQL access to your tenant — connect a BI tool or run ad-hoc queries against your tenant for full analytical flexibility.

Ask the Git AI agent for help connecting to the API or setting up ETL jobs to sync this data into your own warehouse.