Register `tencent/workbuddy-bench` as an official Benchmark dataset on the Hugging Face Hub

Hi team! thanks for open-sourcing the benchmark and the eval harness :)

I noticed the dataset already lives on the Hub at tencent/workbuddy-bench, but it isn't registered as a Hub Benchmark yet. Registering it would upgrade it from "just a dataset" to a benchmark the Hub actively tracks, with a few concrete benefits:

• Dynamic leaderboard: the Hub aggregates evaluation results submitted to any model repo and renders a live, sortable leaderboard on the dataset page, no need to hand-maintain a leaderboard table/site.
• Discoverability: official benchmarks get a "Benchmark" tag and show up in the official benchmarks browser, so people looking for agentic/coding-agent evals can find WorkBuddy Bench without already knowing about it.
• Results travel with the model: once someone evaluates a model and submits a small YAML to that model's repo, the score shows up both on the model page (linking back to your leaderboard) and on your leaderboard.
• Programmatic access: once registered, anyone (including agents/scripts) can pull the leaderboard via HfApi().get_dataset_leaderboard("tencent/workbuddy-bench") or GET /api/datasets/tencent/workbuddy-bench/leaderboard, and per-model results via model_info(..., expand=["evalResults"]). Also embeddable as an iframe on workbuddybench.com if useful.
• You can directly add info about what harness was used for each model, allowing you to compare harnesses on the hub.

Docs:

• Registering a benchmark: https://huggingface.co/docs/hub/eval-results#registering-a-benchmark
• Fetching leaderboard data via API: https://huggingface.co/docs/hub/leaderboard-data-guide

The process is pretty lightweight: add an eval.yaml to the dataset repo root describing the benchmark/tasks (see GPQA, SWE-bench Verified, or HLE's eval.yaml for examples), then we will add you to the allowlist of benchmarks!

Start in the dataset repository root by reading the Hugging Face documentation on registering benchmarks and comparing the eval.yaml examples for GPQA, SWE-bench Verified, and HLE. Add an eval.yaml describing WorkBuddy Bench's benchmark and tasks, then verify that the dataset is accepted for the official benchmark allowlist.