Isn't $f(X)=X^3+X^2-2X-1$ the minimum polynomial of $u=2\cos(2\pi/7)$? If so then $K=\mathbb{Q}(u)$ is a degree $3$ Galois extension of $\mathbb{Q}$.
But a pure cubic field $L=\mathbb{Q}(a^{1/3})$ cannot be Galois, as its Galois closure contains all cubic roots of unity. These are quadratic over $\mathbb {Q}$, so $L$ is not Galois over $\mathbb{Q}$.
What's wrong with Galois theory! Anyway, the field $\mathbb{Q}[X]/\langle f(X)\rangle $ has three maps to $\mathbb R$ taking the image of $X$ to $2\cos(2\pi/7)$, $2\cos(4\pi/7)$ and $2\cos(6\pi/7)$ in the three cases. But $\mathbb{Q}[X]/\langle X^3-a\rangle $ only has one homomorphism to $\mathbb R$ taking $X$ to the unique real cube root of $a$.
How's that?