Anthropic Ships Contribution Metrics for Claude Code Groups

Anthropic launched contribution metrics for Claude Code on Thursday, giving engineering groups a method to measure how a lot the AI coding assistant really impacts their output. The characteristic, now in public beta, integrates straight with GitHub to trace pull requests and commits made with Claude Code’s assist.

The numbers from Anthropic’s inside utilization inform an attention-grabbing story. Since ramping up Claude Code adoption, the corporate says its engineers now merge 67% extra PRs per day. Much more hanging: 70-90% of code throughout groups is now written with Claude Code help.

What the metrics really present

The GitHub integration surfaces three foremost knowledge factors: PRs merged (cut up between assisted and unassisted), strains of code dedicated with and with out Claude Code, and per-user contribution patterns throughout groups.

Anthropic is being cautious about attribution right here. The system solely counts code as “assisted” when there’s excessive confidence Claude Code was concerned, matching session exercise with GitHub commits. Workspace admins can entry all the pieces by the prevailing Claude Code analytics dashboard—no further tooling required.

The productiveness measurement drawback

Pull requests aren’t an ideal proxy for developer productiveness, and Anthropic acknowledges this straight. However they argue PRs correlate intently with what engineering groups really care about: delivery options, squashing bugs, and getting updates to customers sooner.

The corporate positions these metrics as complementary to current engineering KPIs fairly than a alternative. Groups can layer them alongside DORA metrics or dash velocity to identify directional adjustments after adopting Claude Code.

Accessible now for paid tiers

Contribution metrics are rolling out to Claude Staff and Enterprise clients. Setup requires putting in the Claude GitHub App, toggling on GitHub Analytics in admin settings, and authenticating to your group’s GitHub account. Metrics populate robotically as groups use the software.

This launch follows Anthropic’s January 2026 launch of Claude Code 2.1 and Claude Cowork. The agentic coding software, which operates straight in builders’ terminals, can deal with all the pieces from constructing options to resolving merge conflicts autonomously. It maintains challenge consciousness and pulls from exterior sources like Google Drive and Figma through MCP integration.

For groups evaluating AI coding instruments, having concrete productiveness knowledge might assist justify the funding—or reveal whether or not the hype matches actuality.

Picture supply: Shutterstock