Start with the mission. First beyond LEO crewed flight is around the moon, the second is to L2. Better yet, the first manned mission goes to to L2: we haven't been there yet and the potential scientific return much greater.
But at L2 we should plan to do work that's worth sending humans, so before that we send a simple man-tended Farside Station. It might be based on an ATV and Node 4. It would have more robust shielding than a capsule could carry and allow longer stays, and have more capable communications. It could have racks for lab animals, testing the effectiveness of the shielding and providing data on the affect of radiation beyond the Van Allen belts.
It would allow the testing of systems essential for long duration manned flight to deep space objects like asteroids and the Martian moons or surface, but still allow relatively quick return in the case of emergencies.
Once the station was in place, the first of a campaign of landers would go to the lunar farside. They would be able to land rovers and return samples to Farside Station.
At the station, crews would teleoperate rovers on the surface, collect samples sent from the Lunar surface and lab animals for return to earth.
A simple Centaur based cryogenic depot would either be attached or fly in formation, providing operational data for that environment. The landers might be reusable single stage vehicles that carried rovers down and samples up and refueled at the depot.
This mission would be less challenging than a manned Lunar landing or even a Plymouth Rock mission to the easiest NEOs, but still provide significant scientific value. I could get enthusiastic about this mission, certainly more so than a Lunar flyby.
Several expeditions could follow the first. International partners could participate by offering redundant logistic and crewed transport support in exchange for their own use of Farside Station without being in the critical path of the project. Additional equipment could be landed on the surface: Power storage or one or more radioisotope thermoelectric generators to enhance the effectiveness of the landed hardware, a drill for deep sampling, and perhaps prototype equipment to process the regolith for oxygen or buried water ice and a robonaut for more dextrous servicing.
Another valuable near term human space flight project would be a variable gravity habitat in LEO. The habitat itself could be a prototype for a lunar surface habitat berthed at one end of the structure. An expended rocket stage could serve as a counterweight at the other. A module at the axis of rotation could provide despun platforms for visiting spacecraft to dock or berth, and a platform for communications. Deployable masts would connect the modules, with suspension cables taking the primary loads when the station was rotating.
The goal would be to provide artificial gravity ranging from as high as Mars surface gravity of .38 g to Lunar gravity of .18 g, or the even lower gravity of Phobos and Deimos, simply by changing the rate of rotation.
Here is a NASA paper on the related problem of providing artificial gravity on a spacecraft bound for Mars.
The paper describes a system that produce 1 g in the habitat, which would require much longer masts and greater angular momentum for the entire system.
And that shows one example of the value of carrying out this program: we have very little information on the long term affects of environments less than 1 g and more than microgravity on the human body.
We know that the human body is pretty well adapted to Earth's gravity. This is not a supirise. Long term exposure to microgravity, on the other hand, has a lot of bad effects, and the human body takes a long time to recover from them.
We know very little about intermediate cases. For Apollo 17, the last mission, two men spent a little over three days on the Lunar surface, plus another nine days and 11 hours in microgravity. The earlier missions were shorter. That's not long term data.
While that happened the first manned asteroid mission would be prepared, more or less as described by HEFT: a deep space craft with a modest hydrogen stage for departure from L-2 and SEP for deep space delta-V, and capable of reaching more than the very easiest targets. The simple cryogenic depot at Farside Station would be sufficient to support this mission.
Alternatively, manned landers staging at Farside station could visit the infrastructure building up on the Lunar surface.
This series of missions could be launched in several ways. It could use the SLS so beloved by space state congressional representatives and MSFC, but the September 2010 HEFT report suggests that because of the enormous development costs and high fixed costs of SLS, the program would need to be stretched and slowed to meet likely budget limitations.
In that document, the first asteroid mission is delayed until 2031, and the program is still $14 billion over budget.
The launcher alone, not including ground operations and the infrastructure to launch it, consumes half the cost of the program. For a point of comparison, launch costs for NASA’s unmanned spacecraft, and for military satellites and commercial comsats rarely exceed 20% of the program cost.
The same program seems much more affordable using existing rockets and propellant depots.
Here's another compelling use for an L2 outpost: the delivery point for an Asteroid Retrieval and Utilization Mission.