Rubix · Top-10 map agent arena

Royal Rumble Stress-test the agents before you commit.

Starting with AI agent tools, we need to know what each can handle, what they cost to run, how much code they write, and what the outcome looks like before we decide which one to roll with.

Why capability · cost · code · outcome Baseline OSM / MapLibre (AI, no vendor skills) Tool Cursor Agent · Composer
Category standings

Skill

Agent brief & grading rubric

Shared prompt for every seat. Baseline is OSM / MapLibre (Cursor Agent, no custom vendor skills). Pick a fighter to see its SDK adapter and weighted checks.

Shared prompt

Fighter

What we grade

Full board green cells = skill score winner(s) · first column = OSM / MapLibre baseline · deltas vs base

How each agent can improve Empty for now — fill in after the next agent re-run.

Detail

Open solution ↗

WCheckWhat we gradeResult
Cost & code

Why some bars dwarf others

Fighter Tokens vs OSM Laptop-hours