These figures come from real npx whoburnedmore runs: the CLI reads the local usage logs your AI coding agents already keep, totals the tokens by day and model, estimates cost from public model pricing, and submits only those daily aggregate numbers. No prompts, code, or file names ever leave a contributor's machine — see what we collect.
Want the per-tool how-tos behind these numbers? Start with the cross-tool usage guide or the full guide library.
One command ranks your token burn on the board and updates this data.