AI Usage Metrics
ShipLens tracks AI-assisted development patterns by analyzing commit metadata for co-author indicators.
Detection
AI assistance is detected from the co_authors field in commit reports. When a commit includes AI tool attribution (e.g., GitHub Copilot, Claude, GPT), it's flagged as AI-assisted.
Detected Tools
The following AI tools are recognized:
- GitHub Copilot — Detected from co-author tags
- Claude — Detected from Anthropic attribution
- GPT / ChatGPT — Detected from OpenAI attribution
- Other AI tools — Pattern-matched from co-author metadata
Metrics
| Metric | Description |
|---|---|
ai_assisted_commits | Count of commits with AI co-authorship |
ai_percentage | Percentage of total commits that are AI-assisted |
ai_tools | List of detected AI tools with usage counts |
AI Percentage
Rounded to 1 decimal place.
How AI Metrics Are Used
AI metrics are informational only — they are displayed in contributor profiles and team summaries but do not affect scoring.
They help managers understand:
- How much of the team's output involves AI assistance
- Which AI tools are being adopted
- Whether AI usage is increasing or decreasing over time
- How AI-assisted commits compare in quality (via slop detection cross-reference)
TIP
AI usage is not inherently good or bad. High AI usage with low slop scores indicates effective AI-assisted development. High AI usage with high slop scores suggests the AI output needs more human review.