Agent Evaluator `print_detailed_results` output is not readable

[nimanthadilz]

Describe the bug
The output printed by AgentEvaluator class's _print_details method is not readable when the columns have long text (ex: expected_response & actual_response). See screenshots below.

To Reproduce
Steps to reproduce the behavior:

1. Any agent setup (better if agent outputs a long response)
2. Create a test_config.json and an evalset file for the above agent
3. Run the evalset with either adk eval command or using pytest as mentioned here.

If you want a quick way to check this, I have a sample GitHub repo with a simple setup to reproduce this here (Make sure to use reproduce-print-detailed-results git branch).

Expected behavior
A more readable output should be printed when using print_detailed_results as shown in the Screenshots.

Screenshots
Not readable output:

More readable output (after fixing this issue):

Desktop (please complete the following information):

• OS: Windows 11
• Python version: 3.13.2
• ADK version: 1.17.0

Model Information:

• Are you using LiteLLM: No
• Which model is being used: gemini-2.0-flash

[nimanthadilz]

Hi all,
When AgentEvaluator class prints the output, it uses the tabulate python package to format the output as a table (here).

print(tabulate(pd.DataFrame(data), headers="keys", tablefmt="grid"))

The issue here is we are not setting a max column width and the tabulate package is using the longest text length for the column width.
My suggestion is to use the maxcolwidths parameter from tabulate to set a reasonable value like 20.

print(tabulate(pd.DataFrame(data), headers="keys", tablefmt="grid", maxcolwidths=20))

Above suggestion would generate an output as shown in the 2nd screenshot.

@ankursharmas let me know WDYT. I am happy to raise a PR if that's fine.

[surajksharma07]

@nimanthadilz line 456 in agent_evaluator.py is missing the maxcolwidths parameter in the tabulate call, causing the table to expand beyond readable width with long responses.Your proposed fix is spot-on: adding maxcolwidths=30 will wrap the text nicely.
We will look more into it for apply the same fix to cli_eval.py line 272 for consistency.
This is a quick one-line fix that'll make evaluation output much more readable.
Raise a PR and link with this issue.
Thanks!