Grounded in available product and source data
Drip answers a question no AI provider publishes: how much water does your coding agent drink. It reads the token counts from your agent's own logs and turns them into a live estimate.
The meter appears in the Claude Code status line, beside Codex CLI, or as a split-pane meter in your terminal. A browser breakdown shows the maths per model, the uncertainty range, the methodology and the sources behind each coefficient.