agent-expenditure

This Skill addresses the problem of opaque token consumption and unquantifiable coordination overhead in multi-agent parallel scheduling scenarios, which leads to resource waste and uncontrolled costs. In practical development, developers often need to dispatch multiple AI agents simultaneously to handle complex tasks, but lack effective tools to track each agent's token usage. This makes it difficult to determine whether adding more agents actually improves efficiency or merely increases coordination costs.

Usage is straightforward: after completing a parallel agent task, simply run this Skill for analysis. It automatically calculates token usage per agent, identifies coordination overhead such as shared file conflicts, and flags potential waste signals like duplicated work or outputs with disproportionate value relative to consumption. You can then use natural language commands to view detailed waste reports and optimization suggestions.

It is well-suited for teams or individuals who frequently employ multi-agent collaboration, particularly those already using parallel scheduling for tasks like code review, data analysis, or content generation but struggle to quantify their benefits versus costs. For project managers and operations engineers, it also serves as a crucial resource control tool.

We recommend using this tool for review after each parallel task involving four or more agents. Note that it primarily performs post-hoc analysis and is not suitable for single-agent workflows or real-time token budget control—the latter should be handled by dedicated token budget management tools.