$ABCDEFG$ is a regular heptagon. $H = AC \cap BD, K = AD \cap BF$ then show that $CDEFKH$ lie on an allipse.
Five points always define a conic, so I figured that is not hard to see that the conic through $FEDCH$ must be an ellipse due the configuration of the points. I don't have a clue about how to solve it, looks like some projective result which I'm not familiar with. I thought about the reverse of Pascal's theorem (Braikenridge-Maclaurin) but it was to no avail. May be related to Poncelet's theorem.

Here is a proof using the reciprocal of Pascal's theorem:
Let $T$ be the mid point of $AB$. Then $E$ is a symmetry axis of the heptagon, so the lines $ET$, $AD$, $BF$ intersect in $K$. Then in $\Delta KAB$ the angle bisectors $K$, $AHC$, $BG$ intersect in a point, denoted by $S$.
Consider now the hexagon $$ EKFCHD\ . $$ As also shown in the picture, the three relevant intersections $EK\cap CH=S$, $KF\cap HD=B$, and the infinity point of the direction $DE\|FC$ are colinear. (The line $SB$ also passes through this infinity point, since it has the same direction $BSG\|CF\|DE$.)
The converse of Pascal theorem insures now the fact that the mentioned hexagon has vertices on a conic. The conic is an ellipse $\mathcal E$, with one symmetry axis $s$ the common side bisector of $DE$ and $CF$, which contains $A$. It also passes through the reflections w.r.t. the axis $s$,
$\square$
We are done, but let us...
Note: The tangent to $\mathcal E$ in the point $C$ is the heptagon side $BC$. And mirroring w.r.t. $s$ the tangent in $F$ is $GF$. To show this, one may consider again a hexagon like $DCCEKF$, and show that the line through $DC\cap EK$ and $S'=CE\cap DF$ also passes through $B$. (The point $AKD\cap CF$ is also on this line.)