I agree that it can be hard to translate the general idea of a project you're still planning into the terms of a issue tracker. This task can be made easier by understanding that a "Plan for a project" is the same thing as "A rough specification for a project", which in turn is the same thing as "A collection of features".
So, just start with the features you want to implement. You can further link more issues to each of those features as you know what changes you need to make to support those features.
Here's How I like to approach this. Once I've gotten to the point that yes, I really want to start the project, My first action will be to place the "elevator pitch" as the first ticket in the issue tracker. Say:
1: A website that makes straighter bacon!
And almost right away, I can see that I need some basic things just get off the ground
2: Create a django project (is depended on by 1:)
3: Create some database models for bacon (is depended on by 1:)
4: Design An HTTP method for cooking food. (Is depended on by 1:)
And I work with laserlike focus to implement all of the features that I need to create a project that actually does what that first ticket says. Being able to close that first ticket, and seeing how much further I have to go before I can close that ticket, Is a very worthwhile and motivating thing.