As others have explained, solar sails can be used to apply gradual delta-V to an orbit in either direction, to go in or out.
However... that means you can apply delta-V to any object, such as a large rock. It only takes a small amount of it to change a "near miss" into a "planet-destroying hit". We get a few decent near misses every year - see https://en.wikipedia.org/wiki/List_of_asteroid_close_approaches_to_Earth
You don't even need to use the sails to move them. You can deflect them in any number of ways, so long as you can get close to them at matched velocity, early enough - which a solar sail would let you do, if you knew the object's expected path.
So all planets and stations having a defense against such asteroids would be necessary to fulfull your "no weapons of mass destruction". Since... well, a near-earth-shatteringly large mass that's been weaponized is very literally a weapon of mass destruction.
In 2013, NASA estimated it had identified 95% of the potential "planet-killers" of 1km or more (about the same as the dinosaur-killer), and only 10% of those smaller than 300 meters.
Tracking rocks appears easy, but there are many rocks that are visible to earth only by obscuring other things, or on close approaches to the sun. Even among those that have been detected, they aren't constantly tracked; their orbital paths are calculated once they're detected, then they are filed away and forgotten about until that orbit is due to become interesting again.
That changes if there are things which might push them out of those predictable orbits.
Even if 100% of dangerous objects were identified and were continuously tracked... solar sails are essentially mirrors, which makes them black to anything outside a very narrow angle, so they are undetectable other than in their effect on the rock. If the solar sail obscures the rock, or if the rock has been covered in carbon-black, then the rock will not be visible to telescopes either.
So the whole system of falling-rock-detection currently on earth would need to be improved several thousandfold, to detect and track rocks far more constantly and aggressively than our current snapshot-based system. We could no longer rely on objects moving in predictable orbits, and no longer rely on passive image detection; a networked array of telescopes detecting stars being obscured, and using some form of active detection (very long distance radar?) for concealed rocks.
Deflecting a rapidly-moving rock requires a whole lot more force than aiming; dropping a thing on a planet is a lot harder than it oughtta be, because it is likely to just get caught in an orbit instead of spiralling inwards, but still, the person aiming has the significant advantage of lead time. The closer it gets, the more force must be applied to deflect it, and I think the increase scales with 1/distance^2.
There's also the shotgun problem: no matter how many rock-deflecting missiles you have in your armory, your foe only needs to send one more planet-destroying rock than that, in order to destroy your planet. Only one needs to get through.
Have one large visible rock followed by a stream of smaller blackened ones hidden behind it.
Have what appears to be one rock spread apart into many once the planet deploys deflection measures.
Scatter larger high-albedo and smaller low-albedo rocks together in the same "shotgun blast", so that the glare from the shiny ones masks the stealthy ones until it's too late to deflect them.
...and so on.