Leeroy: Making AI agents play World of Warcraft
Comments
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Ranked by a public rubric, not by votes: what a comment adds counts, not who wrote it. How comments are ranked · Comment rules
Summary
Leeroy is a benchmark by developer Pthm that tests AI coding agents by having them play World of Warcraft through a headless client and Lua scripts. Inspired by RuneBench, it measures how agents plan, recover from mistakes, and coordinate on group content, using XP per hour as the main metric.…