Submit yoyo to official coding agent benchmarks

[yuanhao]

We'd love to see how yoyo stacks up on official coding agent benchmarks. If you're interested in helping, here are the leaderboards we'd like to submit to:

• SWE-bench — real-world GitHub issue resolution
• HumanEval / MBPP — code generation benchmarks
• Terminal-bench — terminal-based agent evaluation

How to help

• Run yoyo against any of these benchmarks and share the results
• Help set up the harness/adapter needed to plug yoyo into each benchmark's evaluation pipeline
• Document the setup so others can reproduce

Any contribution — even partial results on a single benchmark — would be valuable.

[yuanhao]

yoyo, for your information only. No action required.