agent-cost-attribution is a stdlib-only Python package that delivers per-stage token and cost attribution along with silent-degradation detection for multi-agent AI workflow runs. It helps developers monitor and optimize the economics and reliability of complex LLM-based agent systems.
In the Agent monitoring & governance space, agent-cost-attribution takes a focused approach. It focuses on tracking per-stage costs, tokens, and performance degradation in multi-agent LLM workflows. agent-cost-attribution is an open-source project aimed at ai developers. The project is open source (MIT). agent-cost-attribution is available on the command line.
Behind agent-cost-attribution is Jott2121, and it first shipped in 2026. Development happens publicly on GitHub with 13 commits in the last 90 days. Among its 3 catalogued features are token attribution, cost tracking, and silent degradation detection.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do