Use fuzzy searching for node list
The current search algorithm seems like a basic substring search. Where your query must contain a single continuous substring of the name you want to match. For example "Flood" will match all the various Flood Fill nodes, but to match a specific node such as "Flood Fill to Position" you have to type "Flood Fill to P".
In many modern software applications a fuzzy searching algorithm is used. Where search results are sorted by the "distance" between two strings where the distance is determined by the number of characters that must be added/deleted/substituted to make a query string match a target string. For example "Flood P" requires the addition of 15 characters - "Fill to osition" - to match "Flood Fill to Position" perfectly.
There are many benefits to this approach. Since the distance is numerical a queries results can be sorted based on how closely they match a given string. This not only allows for even more fragmented queries to match targets more accurately, such as allowing a user to type initials "fftp" or short fragments "fltopos" to accurately match a desired target of "Flood Fill to Position". Depending on how completely an implementation attempts to transform the query string in to a target string these algorithms can even handle spelling errors - though in the real time context of search menu being this thorough can have performance issues.
Beyond the improved UX experience of being able to use shorthand queries. There's also how it results in a much more beginner friendly experience. You know you want a "flood fill" node but can't quite remember what it was called, was it position or displacement or warp? Just type enough to match "flood fill" such as "flood", read the query results and refine your query with the results as they update, e.g. "flood pos".
A popular MIT licensed implementation of the Levenshtein Distance search is FuzzyWuzzyMIT. Which has been ported to several languages.