Test Cadence's AI detection capabilities on your own repositories and websites.
Common questions about the analysis playground
Cadence uses 38 detection strategies to identify patterns commonly associated with AI-generated content. While highly effective at spotting suspicious patterns, it's not 100% accurate. Results should be used as one factor in your evaluation, not the sole determining factor.
You can analyze any public GitHub or GitLab repository by providing its URL. Private repositories are not currently supported in this playground. For private repository analysis, consider running Cadence locally or deploying your own instance.
Analysis time varies based on repository size and commit history. Small repositories typically complete in under a minute, while larger repositories with thousands of commits may take several minutes. The playground will show real-time progress as your analysis runs.
A high suspicion score indicates that Cadence detected multiple patterns commonly associated with AI-generated content. This could include rapid commit velocity, unusual timing patterns, generic commit messages, or structural anomalies. However, legitimate code can sometimes trigger these patterns too.
This playground is designed for testing and evaluation purposes. For production use cases, we recommend running Cadence locally with your own infrastructure, reviewing the detailed analysis reports, and combining Cadence results with other evaluation methods and human review.
Analysis results are processed in real-time and stored temporarily during the job execution. We do not permanently store your repository data or analysis results. However, URLs you submit may be logged for debugging and rate-limiting purposes.