Every time somebody asks a question like "can I use X spell for doing Y" the answer is usually "no" because spell descriptions are very short in 5e, and usually they don't explicitly say a spell can do Y.
These answers are based on the "spells do only what they say they do, nothing more" principle.
What is the source of this premise?
Related question: Is there a rule for how to handle creative use of spells?
I'm asking this as a DM. My first thought was "well it is obvious, why should spells do something more". But the more I dig into this topic, the more contradictory arguments I find. I've gathered all thoughts down below, if anyone is interested (upvoted comments indicate that people are).
DMG examples
The Dungeon Master's Guide has examples of spells doing things out of their originally described scopes:
An area of desecrated ground can be any size, and a detect evil and good spell cast within range reveals its presence. (p.110)
An identify spell reveals that a creature is inside the flask (p.178)
one torch can burn a Huge tapestry, and an earthquake spell can reduce a colossus to rubble (p.247)
Other factors might help or hinder the quarry's ability to escape, at your discretion. For example, a quarry with a faerie fire spell cast on it might have disadvantage on checks made to escape (p.253)
RPG.SE answers
There are highly upvoted answers, implying creative spell usage is a thing in 5e:
Your players are using spells creatively…
That is exactly what D&D 5 encourages.
Open-ended spell descriptions
Some spells have quite open-ended wordings, like Prestidigitation:
harmless sensory effect, such as a shower of sparks, a puff of wind …
Other spells descriptions aren't very detailed in 5e (compared to systems like Pathfinder), I guess that means it is the DM's job to ultimately say what happens when somebody use a spell in an unusual way.
DMG supports this with its common principle, in Chapter 8: Running the Game (page 235):
Rules enable you and your players to have fun at the table. The rules serve you, not vice versa.
Trusted third-party sources
Some third-party sources encourages stretching spell limitations. See Geek&Sundry "How Watching Critical Role Made Me Better At D&D":
A spell is typically written vague enough that you don’t have to worry about specific limitations unless you’re trying to stretch them. When in doubt, explain to the DM what you want to do and see if they’d be game
See also The Rule of Cool by Matthew Mercer.
Common sense
The strict "spells never do anything their description doesn't mention" principle simply doesn't work. When a player asks "Is Grease flammable?" they already challenge the frame, regardless of the answer. If the DM says "yes", you can ignite the grease. If the DM says "no", you can extinguish flames using the grease.
Of course, since "spells do only what they say they do", DM might say "no" to both assumptions, but this effectively turns a tabletop role-playing game into a pen and paper computer game, boring and awkward. I don't think this is actually RAI.
It seems a DM is supposed to resolve an unusual spell application case, using the intent behind the spell, rather its literal description. For example, it seems reasonable you should be able to use any fire-producing spell to light a torch in a non-combat situation, even when its description doesn't explicitly say that it "ignites flammable things". It heavily depends on the particular DM's style, like many other things in 5e.
What is the source of the "spells do only what they say they do, nothing more" principle?
Best Answer
The rule is an extension of a precedent set by WoTC themselves
The lead rules designer of 5e, Jeremy Crawford, has1 the power to make official rulings, and frequently does so on Twitter, and in the Sage Advice column on the official D&D site. It's common for him to answer questions with some variation of "if a feature was meant to work that way, it would say so." He has even explicitly stated:
Using this principle, he has made rulings such as:
or
From this, we can derive that barring some explicit clarification from Sage Advice, JC himself, an official errata, or a more specific rule mentioned somewhere else in the game's official material, features in the game are intended to only do what they say. Though, of course, the DM is permitted to make their own rulings and allow spells and effects to do things not directly stated in their description as they see fit.
1 While this status was subsequently backed away from (official rulings only now apply to the Sage Advice Compendium as posted by WoTC, not the tweets from J Crawford), it was true that his ruling was authoritative when this question was asked, and answered; that change in state does not change the overall point.