tokentab validates demand for agent cost accounting

AI agents are becoming a real operating cost, especially for software teams using several coding tools at once. The issue is simple. A business can see its overall AI bill, but it often can’t tell which client project, product team, or piece of work created that spend.

That is the gap tokentab is aiming at. It’s a command-line tool that tracks token spend by project across three coding agents. In its first five days, it picked up 527 GitHub stars, according to the project’s GitHub repository. Stars are not revenue, but they are a useful early signal that developers and operators want clearer visibility into where agent costs are going.

For firms delivering AI work to clients, this matters commercially. An AI-ops retainer can’t stay vague for long. Clients will ask what they are paying for, which workflows consume the budget, and whether the output justifies the cost. Teams that can attribute AI usage to specific projects can price work with more confidence, spot margin leaks earlier, and have better conversations when usage spikes.

This is the kind of thing we build into an AI command centre, bringing usage, cost, activity, and outcomes into one view so operators can manage AI as part of the business, not as an untracked expense.