AI Code Insights
AI coding agents are writing code across your organization every day. AI Code Insights shows you what they produce, how they work, and where they get stuck — so you can measure AI’s real contribution and unblock agents that aren’t delivering.
Capabilities
AI Code Insights lets you:
- See which commits and PRs contain AI-generated code, broken down by team, repository, agent, and time period.
- Collect session-level data from supported agents — transcripts, summaries, and token usage — to understand not just whether agents are being used, but how.
- Trace AI coding sessions to the commits, PRs, and deployments they produced.
- Use Agent Experience to surface insights about where AI agents are hitting bottlenecks, evaluated from the agent’s perspective.
How it works
- Install the AI Code Insights daemon
- Work with a supported agent
- Commit and push your changes
AI Code Insights attributes AI code when you commit, and pulls in per-session details as sessions end. The daemon monitors repositories whose Git remote origin matches a repository you have syncing to DX.
Supported agents
| Agent | Code attribution | Transcripts | Token usage | Session output |
|---|---|---|---|---|
| Claude Code | Yes | Yes | Yes | Yes |
| Cursor | Yes | Yes | No | Yes |
| GitHub Copilot | Yes | Yes | No | Yes |
| Codex | Yes* | Yes | Yes | No |
| Gemini CLI | Yes | Yes | Yes | Yes |
| OpenCode | Yes | Yes | Yes | Yes |
| All other AI coding agents | Yes* | No | No | No |
* Code attribution via behavioral analysis (file system observation) rather than a native hook. Session output requires a native hook or plugin.
Available reports
- AI code percentage — AI-generated code percentage across teams, repos, and agents
- AI pull request overview — Pull request metrics broken down by percent of AI code per PR
- Agent Experience — Per-session scoring of agent friction and effectiveness
- AI Dollar Impact — Estimated dollar value of AI agent usage
- Data Studio — Custom queries and dashboards over your AI Code Insights data
Privacy
AI Code Insights is designed to measure code authorship, not monitor developers. No source code leaves the developer’s machine, transcripts are scrubbed for secrets and PII before transmission, and admins control exactly what data is collected. See Security and privacy for the full breakdown of data flow, what is and isn’t collected, and endpoint considerations.
Get started
See Installation to get started, MDM installation for organization-wide deployment, or Rollout for pilot strategy.