Evaluate the agent's output against the criteria below and give it a score from 1 to 5.

<criteria>
$criteria
</criteria>

<input>
$input
</input>

<output>
$output
</output>
$reference
Scoring scale (use the whole scale):
5 - fully meets every criterion; nothing important is missing or wrong.
4 - meets the criteria with only minor issues.
3 - partially meets the criteria; a noticeable part is missing or wrong.
2 - mostly fails the criteria; only small parts are acceptable.
1 - does not meet the criteria at all, or is wrong, harmful or off-topic.

First explain your judgement in one to three sentences, citing the output, then give the score.
Reply with only this JSON object:
{"reasoning": "<your explanation>", "score": <integer from 1 to 5>}
