# What is cause attribution?

Cause attribution is the step from knowing a regression happened to naming the specific change that caused it. Detection establishes that quality dropped; attribution is the investigation that follows, over a finite set of candidates: a code commit, a prompt edit, a model change, a tool whose behavior shifted, or an upstream system the agent depends on.

The investigation narrows that set against evidence. You gather the failing traces into one cohort, line their onset up against the change history, and test each candidate until the ones that don't hold are ruled out. Failures cluster around the code path that produces them, so when the failing traces share one path, the candidate set shrinks from everything shipped recently to the handful of changes that touched that path in that window.

A finished attribution names the cause, cites the traces and findings that support it, and lists what was checked and ruled out, which is also what tells you where the fix belongs.

---

Source: https://tessary.ai/answers/cause-attribution/what-is-cause-attribution
More on Cause attribution: https://tessary.ai/answers/cause-attribution
From Tessary, agent reliability for AI agents in production: https://tessary.ai