Ahhh, but what’s in the current episode object? The pointer knows but it's keeping it on the down-low... 
json_select does do this already for key:values, json_select_object as you rightly say describes it but doesn't actually output the contents until probed by other commands. Sometimes it's far easier to have full sections available as actual text for commands other than json-based ones to access. I'll even risk an assumption this could be made to help HTML parsing.
Examples are at the bottom but for more context, I’ll attach an example query response from my TMDB scripts:
AlbumInput.txt (129.4 KB)
This is everything necessary for "Game of Thrones" season 1. Part of the script’s speed is avoiding iterations where possible and using methods to select and process multiple batches of data at once using RegexpReplace, like getting all the actors at once in just a few commands. A simpler example is checking if an episode is available:
Use "%INPUT%"
RegexpReplace ".*?\"episode_number\":(%NUMBER%),.*" "$1"
From testing other movies/TV, the response data can also include data with newlines (\n), quotations (\”) and vertical bars (|) needing replaced as RegexpReplace doesn't like these. That means having to json “on” “current” the transformed response inside an output buffer as opposed to json “on” whenever a new section of data is required in text form.
Just Useing this “Input” buffer quickly ramps up a debug file's file size and there’ll be times when I need tons of data processed to find a fault within a debug file foregoing the file size parameter and hammering CTRL+F with any clues to the cause. Even with careful planning to stop or reduce an impact, I’ve often accidentally had debug files nearing 1GB... which to clarify is the disgusting part. 
The ability to reveal currently pointed-at sections as “raw” text through this suggested command would greatly reduce this. It could also save some extra prep-work transforming strings of data that would otherwise be readily usable for commands to use.
I admit I’ve still a way to go to become more professional with debugging strategies though I do use movies/TV that return smaller, controlled responses and try to use separate test scripts for each new module of code. I also use this and that to validate and proof-read (but your site does look better 
) as it is vital to review the debug files no matter how nasty.
So with a progressing example of a potential command:
After using json "on", reveal as text:
...129.4 KB's worth of text.
After json_select_object "genres":
[{"id":10765,"name":"Sci-Fi & Fantasy"},{"id":18,"name":"Drama"},{"id":10759,"name":"Action & Adventure"}]
After json_foreach "genres", it'll be 1/3:
{"id":10765,"name":"Sci-Fi & Fantasy"}
After json_foreach_end, 2/3:
{"id":18,"name":"Drama"}
And as a final example, after json_foreach "" on ["one","two"] at 1/2:
one
Edit: Okay, one more example.
IndexInput.txt (36.8 KB)
Assuming from a smaller selection of the set:
{"certifications":{"DE":[{"certification":"12"}],"BR":[{"certification":"14"}]}}
...transformed into an array for json_foreach or json_select_array to be able to use:
{"certifications":[{"DE":[{"certification":"12"}]},{"BR":[{"certification":"14"}]}]}
I'd want to get each region (DE,BR) from each object in the array to then get its own contents but given this list could update as new regions are added/removed, I wouldn't know every possible value to json_select. I would need to read the the region key from each object somehow. Currently, it's cross-referencing the loop of objects with a separate array of all region lifted using regex...
OR after json_foreach_end, 1/2:
{"DE":[{"certification":"12"}]}
with a quick regex to lift the region key, use it to access its own contents, use it to name its own contents for outputs, sorted. Either that or again, ways for other json commands to read unknown keys and/or output such otherwise inaccessible keys.
Basically, it could make debugging easier, expand more JSON functionality and give commands more accessibility to the response.
)
very much 
), it would be great to have a command that reveals in raw JSON text what’s currently being pointed at: a single key:value, an object, a 
)