Monitor Cursor AI Spend to Prevent Costly Oversights

This tool addresses the growing concern of uncontrolled spending on AI tools within organizations. The creator's experience of a developer incurring $1,200 in a single day due to using a more expensive model highlights the potential for significant cost overruns. To mitigate this risk, the tool connects to Cursor's Enterprise APIs and provides a comprehensive breakdown of AI spending on a per-developer basis, categorized by model. It incorporates a three-layer anomaly detection system, utilizing thresholds, z-score analysis, and trend analysis to identify unusual spending patterns. When spend spikes are detected, the tool sends alerts via Slack or email, enabling prompt intervention.

In addition to monitoring and alerting, the tool includes incident tracking capabilities, measuring Mean Time To Detect (MTTD), Mean Time To Investigate (MTTI), and Mean Time To Resolve (MTTR). This allows organizations to track the effectiveness of their cost management efforts and identify areas for improvement. The tool is self-hosted, uses SQLite for data storage, and is packaged as a Docker container, making it relatively easy to deploy. It is also MIT licensed, making it open-source and freely available for use and modification.

The tool's ability to save a team approximately $6,000 per month by identifying developers running long sessions with massive context caches demonstrates its potential value. By providing visibility and control over AI spending, it empowers organizations to optimize their usage of AI tools and avoid costly oversights. However, its effectiveness depends on the accuracy of the underlying data from Cursor's Enterprise APIs and the proper configuration of anomaly detection thresholds.