Antigravity Usage Intelligence: Track Tokens, Quotas, Cache, Reasoning, and Agent Activity

Antigravity Usage Intelligence is an open-source tool designed to help developers track and understand their usage of the Antigravity AI development assistant. Created by Nir-Bhay, the project addresses the challenge of determining how much of the AI's work is actually being used, as the agent can perform multiple turns, call tools, read files, generate output, use reasoning, and reuse cached context during a single task.

The Antigravity Usage Intelligence dashboard breaks down usage into several parts: fresh input tokens, cached prompt tokens, output tokens, and reasoning or thinking tokens. This distinction is crucial because two agent sessions may both appear to be large workloads, yet their token behavior may differ significantly due to reliance on cached context.

Key features of the Antigravity Usage Intelligence include live quota monitoring, live token trends, model-level visibility, reasoning tokens visibility, and session-level inspection. The dashboard provides a visual view of when you were using Antigravity, helping to identify heavy coding days, quiet periods, long runs of activity, and changes in usage patterns. Users can also see which models are being used and track tool requests, such as reading files, searching code, running commands, and writing files.

Privacy is a significant aspect of the project, as it operates locally and reads data from the Antigravity assistant without sending usage history to a remote analytics service. The project relies on loopback communication with the local language server, ensuring that active sessions are not disturbed, and there are no external telemetry calls.