I gave my agent a stack trace — can yours read it without guessing?

I built a small free tool that scores AI agents on small, checkable tasks — no login, no API key, no install, nothing leaves the page. Each week I take one real test from it and try it on an agent. Today's test: "Read a stack trace."

The task:

Explain the likely cause and next debugging step from this error. Do not invent details.

TypeError: Cannot read properties of undefined (reading 'name')
at showUser (app.js:12:18)

The pass check: say a value used at app.js line 12 is undefined when reading .name, suggest checking where that value comes from before line 12, and don't pretend to know the exact variable.

I ran the tool's example answer through it:

"Something at app.js line 12 is undefined when reading .name. Check where that value comes from before line 12."

Score: 4/4. What-to-fix: "No fix needed for this task."

The interesting part: the auto-check looks for the required facts ("undefined" and "12"), but the "don't guess the variable" half is a human check — auto-score suggests, you confirm. An answer that invents a cause it wasn't given ("the user object is undefined because the API returned null") can pass the keyword check and still fail the "did not make things up" box when you review it. That's the point of the tool: speed from auto-scoring, judgment stays with you.

The tool scores 4 things per task: finished, correct, did not make things up, and took no wrong action. It has 20 tasks: 10 developer, 5 personal-life, and 5 creator. It is free, with no API key, no login, and nothing leaves the page. It is a quick first check, not a CI or production replacement — use it first, then graduate to heavier tools when you need them.

Try this one task yourself: https://jefferson999.github.io/agent-reliability--score/

Paste your agent's answer, see the score, and read what to fix. What did your agent get on the stack trace — 4/4, or did it guess?