Let
$$\Phi_n(x) = \prod_{0<k\leq n, \gcd(k,n)=1}(x-e^{\frac{2\pi i k}{n}})$$
be the $n$-th cyclotomic polynomial. By observation it seems that cyclotomic polynomials when evaluated at $x=-1$ give the following
$$\Phi_n(-1) = \begin{cases} 2 \quad \text{if} \quad n = 2^k, k>1\in \mathbb{N}\\ p \quad \text{if} \quad n = 2p\\ 1 \quad \text{otherwise} \end{cases}$$
where $p$ denotes a prime $(> 2)$. The first few cases are $$\Phi_1(-1) = -2$$ $$\Phi_2(-1) = 0$$ $$\Phi_3(-1) = 1$$ $$\Phi_4(-1) = 2$$ $$\Phi_5(-1) = 1$$ $$\Phi_6(-1) = 3$$ $$\Phi_7(-1) = 1$$ $$\Phi_8(-1) = 2$$ however we will only consider $n>2$ for the conjecture.
This I thought was quite weird at first as it seems like a very simple pattern for something as complicated as the coefficients of cyclotomic polynomials. I managed to prove the first two cases, and will share the short proofs below, but am unsure of how to tackle the last case. I don't have any way to check for high values of $n$ other than calculation, and so I apologise in advance if there is a counterexample to this conjecture that can easily be checked with a computer. Below are the proofs for $n= 2^k$ and $n=2p$,
$n = 2^k$
By Möbius inversion we get $$\Phi_n(x) = \prod_{d\mid n}(x^d-1)^{\mu\left(\frac{n}{d}\right)}$$
Since $\mu(2^n) = 0$ for $n>1$ with $\mu(2)=-1$, therefore $$\Phi_{2^n}(x) = (x^{2^n}-1)(x^{2^{n-1}}-1)^{-1} = x^{2^{n-1}}+1$$ $$\Phi_{2^n}(-1) = 2$$
$n = 2p$
We can use the following identity $$\Phi_{2p}(x) = \sum_{k=0}^{p-1}(-1)^kx^k$$ $$ = \sum_{k=0}^{p-1}1 = p$$
But for the 'otherwise' case, I don't know how to approach it. It seems to general to approach in any specific way using identities or by looking at the values of $\mu(d)$ as that would require more information about the divisors of $n$. Any help or a proof/counterexample is appreciated!
We use the notation $ f(x) \mid_{x = a } $ to denote we are evaluating $f(x)$ at $ x= a$. This allows me to list out $f(x)$, then evaluate it.
All that we use is $ \prod _{d\mid n } \Phi_d(x) = x^n - 1$.
$$-2 = x^n - 1 \mid _{x=-1} = \Phi_1(-1) \times \Phi_n (-1) \times \prod_{d \mid n, d\neq 1, d\neq n} \Phi_d(-1) = (-2) \times \Phi_n(-1) \times 1 .$$
When naively extending this argument to even $n$, we run into an issue where $\Phi_2(-1) = 0 $ and $ x^{n} - 1 = 0 $, so we can't determine what $\Phi_n(-1)$ will be.
To deal with this, we account explicitly for those zeros via $$ \frac{ x^{n} - 1 } {x +1 } \mid_{x=-1} = \prod_{d \mid n, d \neq 2 } \Phi_{d} ( - 1 ).$$ So, to get to $ \Phi_n(-1)$, we have to evaluate the LHS.
Let $ n = 2^k m$, where $m$ is odd. Then,
$$ \begin{array} {l l ll } \frac{ x^{n} - 1 } { x + 1 }\mid_{x=-1} & = \frac{ x^n - 1 } { x^{2^k} - 1 }\mid_{x=-1} & \times & \frac{ x^{2^k} - 1 } { x + 1 }\mid_{x=-1} \\ & = (x^{2^k(m-1) } + x^{2^k(m-2) } + \ldots + x^{2^k } + 1 ) & \times & (x-1)(1+x^2)(1+x^4) \ldots ( 1 + x^{2^{k-1} } ) \\ & = m & \times & -2^{k} \\ & = - n \\ \end{array}$$
Corollaries:
Quick observation that since $\frac{ x^{n} - 1 } {x +1 } \mid_{x=-1}$ is non-zero, hence so are the individual terms $\prod_{d \mid n, d \neq 2 } \Phi_{d} ( - 1 )$. (This should be obvious beforehand.) Thus, we don't have any more issues to deal with, and can iteratively calculate $ \Phi_n(-1)$.
$\Phi_{2^k} ( -1) = 2 $ for $ k > 1 $
$ \Phi_{2p^k} (-1) = p$ for odd prime $ p$
$\Phi_{2n} (-1) = 1 $ for any non-prime power $ n > 1$.