Token replacement replaces non-token substring like [0]

The token replace function has an option clear = TRUE and within ECA we call the replace function with that clear option being set most of the time. So, a token that's not found is being removed from the string. What you're describing is happening on purpose.

A workaround may be to replace the [0] substring with some placeholder, e.g. ###ph###, and then, after the token replacement, do a search and replace to switch that back to [0].

Thanks, I've tried but the search and replace result shoud be stored in a temporary custom token to be used as url for the redirect action.
In that last action, I shoud use 'replace tokens' to be able to use my custom token and then [0] (token in token) is still removed from the string...

It sounds like you can turn the order around: first replace all your custom tokens and finally s&r the placeholder. As a result, you get a custom token that contains the fully resolved URL.

Couldn't find a way to avoid [0] to be replaced as a token at one step or another so I ended encoding the url so /search?f[0]=tags:[term:id] is written in ECA as /search?f%5B0%5D=tags%3A[term:id] and it worked.
Thanks @jurgenhaas anyway for your kind support for an non-related ECA issue.