Analyze AI coding agents’ work

AI coding agents create pull requests end-to-end, not just suggest code in your editor — they plan work, write code, run tests, and respond to reviews. Swarmia’s new Coding agents view helps you see how these agents are contributing across your organization.

• Agent adoption: See PRs created by GitHub Copilot, Cursor, and Claude Code across all your teams and repositories.
• Usage patterns:
  • Track how many agent PRs get merged versus closed to understand whether they’re mostly throwaway experiments or actually end up in production.
  • Follow the share of PRs created entirely by agents to judge how often engineers need to intervene manually.
  • Examine the batch size distribution to understand the complexity of the tasks the agents are able to complete.
• Trends over time: Understand whether agent usage is growing, and what’s driving it.
• Team comparisons: See which teams are getting the most value from agents.
• Drill-downs: Look at individual PRs to understand what’s working and what isn’t.

Whether you’re in leadership looking to understand adoption patterns across your organization, or on a team trying to figure out what works and what doesn’t, this view gives you the answers you’re looking for.

To get started, navigate to AI tools → Coding agents. The view will work out-of-the-box, with no need for you to configure anything.