Open Issues Need Help
View All on GitHub CONTRIBUTING + adapter/task authoring guide about 2 hours ago
documentation good first issue roadmap
Wire up a Harbor benchmark adapter and run your coding agent side-by-side with industry-leading agents. Arena adds regression checks so your agent keeps improving instead of drifting or degrading, and shows — fairly and without bias — how it stacks up against Claude Code, Gemini, and other top code agents.
TypeScript