Skip to content
The Chat Archive
vicuna-7b Arena Human Preference 55K

give 20 random D&D 5e style magic items you might find in a dungeon that has goblins and…

  1. User

    give 20 random D&D 5e style magic items you might find in a dungeon that has goblins and a wizard

  2. Assistant
    1. Amulet of the Goblin King: This amulet, once worn by the goblin king himself, grants the wearer a +2 bonus to Intimidation checks and a 10% chance to stun any goblin that looks at the wearer.
    2. Wand of Fireballs: This wand can cast fireball once per day, dealing 15d6 fire damage in a 15-foot radius.
    3. Ring of Protection: This ring provides a +2 bonus to AC against goblin attacks and a 50% chance to reflect any damage dealt by goblins back to them.
    4. Boots of Speed: These boots allow the wearer to move at a speed of 30 feet and double their movement on any d20 roll.
    5. Cloak of Invisibility: This cloak grants the wearer the ability to become invisible for 1 minute, once per day.
    6. Lute of Repose: This lute can fill a large room with calming music, putting any goblins in the area to sleep for 1 hour.
    7. Rod of Lightning: This rod can cast lightning bolt once per day, dealing 12d6 lightning damage.
    8. Shield of Faith: This shield provides a +2 bonus to AC and can absorb 5 points of damage per day.
    9. Sword of the Wild: This sword allows the wielder to communicate with animals and gain a +2 bonus to Survival checks.
    10. Potion of Healing: This potion can restore 20 points of health.
    11. Wand of Magic Missiles: This wand can cast magic missile once per day, dealing 1d4+1 force damage per missile.
    12. Ring of Invisibility: This ring grants the wearer the ability to become invisible for 1 minute, once per day.
    13. Boots of Levitation: These boots allow the wearer to levitate for 1 hour per day.
    14. Gauntlets of Ogre Power: These gloves grant the wearer a +4 bonus to strength and a 20% chance to stun any creature they hit with a melee attack.
    15. Wand of Slowing: This

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.