Fix it live

Change the prompt, run the real model, pass the check.

Now fix them for real. Change the system prompt or a tool description, run the agent live on your own free Gemini key, and see if the check passes.

Agent LabGoal: When nothing turns up, say so. No invented number.

System prompt

Tool descriptionssearch

ModelMax model calls

Free key: plenty of runs a day

Your Gemini keyGet one free Stays in this browser and goes only to Google. Never to us.

0 calls today

The recorded run, before your changes:

Searching for something that isn't theregemini-3-flash-previewRecorded real run

Task What was the exact attendance at the first ever IPL match?

You

task

Model1.3s

Model call 1. It read 75 tokens and wrote 18. It asked for a tool, so it isn't done.

search({"query":"first ever IPL match attendance"})

stop_reason: tool_use

Context window

75 tokens sent

0.01% of the 10,48,576-token window. Every call sends all of it again.

Context window75 tokens sent