According to LMFDB the integral points of the elliptic curve $$y^2+y=x^3-3x+4$$ are $(−2,1)$, $ \left(-1, 2\right) $, $ \left(1, 1\right) $, $ \left(2, 2\right) $, $ \left(4, 7\right) $, $ \left(13, 46\right) $, $ \left(19, 82\right)$.
But I found that: $(x,y)=(-2,-2),(-1,-3),(1,-2),(2,-3),(4,-8),(13,-47),(19,-83)$ are also solutions, but they are not listed. Why is that?

This could not be a comment, so it is an answer... The short answer is:
https://www.lmfdb.org/EllipticCurve/Q/135/a/1
explains that only one of $\pm P$ is shown.
Long answer: I am trying to explain the reason by using sage. Here, if we ask for the integral points...
we get the above list of seven points. Since most people most times ask for the integral points on a curve given by the simpler equation $y^2=x^3+ax+b$, and in this case we obviously see the points coming in pairs $(x,\pm y)$, there is a default implicit decision in the computation to show only "the half" of the points. To see them all, we need to explicitly ask for
and we get more. In this case, for a specified value of $x$, say $x=x_0=19$ we know, that the equation in $y$ $$ y^2+y = (19^3-3\cdot 19+4) $$ has an integer solution, the other one is (Vieta) making their sum equal to $-1$.
There is often a reason to "compute / show only half of the points", databases need half space, computation algorithms may make a choice...
And indeed, let us ask sage to show the pairs of points $P, -P$ in the same line: