Add the following bullet points to the character race write-ups provided on p. 8 of the rulebook:
1 free skill level in either Enchanter or Warrior.
1 free skill level in either Scout or Spellcaster.
1 free skill level in either Scholar or Thief.
1 free skill level in any skill.
Chosen during Step 2 ("Select Race") of character creation (p. 5), each of these free skill levels comes in addition to the one provided in Step 3 ("Select Skill"). However, characters cannot double up in a given skill before play begins and Development Points are earned. The free skill level provided in Step 3 must therefore be taken in a different skill than the one selected in Step 2.
Examples: Mike wants to play an Elven mage, so he takes a level of Spellcaster during Step 2. In Step 3, he cannot raise Spellcaster to level 2 but must instead take a level in another skill. He chooses Warrior. Bill wants to play a Human aristocrat, so he selects Leader as his free racial skill during Step 2 and then Scholar as his free skill in Step 3.
I developed this house rule in response to my players' concerns about a bottleneck in character creation. Because five of the game's eight skills can only be used with training, any character who wanted one of those skills had to devote their free skill level to it. The player who chose Spellcaster benefited from this bottleneck since his trained skill was also his primary attack skill; the player who chose Cleric was able to use his Smite ability to similar effect (when wielding one's deity's weapon of choice, Cleric skill may be used in place of Warrior skill). But the players of the Dwarf Scholar and the Human Leader weren't so lucky: they had to spend their free level in Scholar and Leader in order to be able to use any of those skills' sub-abilities and thus ended up whiffing their attack rolls nearly 66% of the time. No one put a level into Enchanter (since the game was a one-shot, not the Enchanter's preferred venue), but I suspect that beginning Enchanters are in the same boat as Leaders and Scholars.
By adding an extra skill level, character concepts involving Enchanters, Leaders, and Scholars are much more viable at the start of play. I decided to attach the extra level to Step 2 and character race to heighten the importance of archetypal origins. Dwarves' choice between Enchanter and Warrior and Elves' choice between Scout and Spellcaster pay homage to the races' folkloric traditions. Thief does much the same for Halflings, looking back to Bilbo Baggins in The Hobbit; Scholar owes more to the Fourth Edition Halfling and that race's tradition of wandering. Humans got a second free choice (to go along with the one they got in Step 3) out of deference to the longstanding RPG tradition of human versatility. Because characters of all races still get the free choice in Step 3, however, no one is locked into a racial archetype. I.e., you don't leave the underkingdom without learning something about either enchanting or fighting—but you can also focus on leading or studying or stealing as well.
There's an obvious counter-argument here: beginning characters are supposed to whiff attack rolls much of the time. While I recognize that argument's force, I felt that the bottleneck didn't distribute the whiffing fairly across characters. Players who devoted skill levels to Cleric and Spellcaster were able to focus their efforts and give themselves 66% success rates (by combining the +10 from the free skill level with the +20 from the primary skill bonus); players who spent their skill level in the three remaining skills had to fall back on their untrained Warrior skill in combat. Adding the free skill level in Step 2 evens the playing field, especially once the prohibition against raising any skill to level 2 before play is considered. Finally, it only costs 3 Development Points (half of the average per-session reward) to buy a skill at level 1 so the house rule really only speeds up the characters' improvement by two hours' worth of play.