If a web source is a well-formed xml document, then you must set the content attribute to xml: this way Curiosity won't apply Tidy before scraping that source.
<webSource name="aSource"
content="xml">
...
</webSource>
Curiosity will assume that the documents retrieved during the scraping steps will have xml content too.
If a screping step apply to a document which has a different content wrt the one specified for the whole web source, you need to properly set the content attribute for that step:
<webSource name="aSource"
content="xml">
<nextstepsList>
<slot level="1"
content="html">
...
</slot>
...
</nextstepsList>
...
</webSource>
Of course, it may also happen that the web source has html content (which is the default) but some steps apply to xml content:
<webSource
name="aSource">
<nextstepsList>
<slot level="1"
content="xml">
...
</slot>
...
</nextstepsList>
...
</webSource>