What problem does it solve? Adding a self-managing ML pipeline (an AutoML framework, LLM agent, or hosted prediction API) to the TabArena benchmark requires creating the right folder structure, subclassing the correct base class, declaring metadata and tags correctly, and wiring up registry and packaging entries. This Skill guides that entire integration so the new system is discovered, classified, and benchmarked correctly. ## Core Features & Use Cases - System scaffolding: Creates the per-system folder under packages/tabarena/src/tabarena/systems/<system_key>/ with system.py (an ExternalSystemModel subclass), hpo.py (a SystemConfigGenerator), and info.py (SystemInfo plus MethodMetadata). - Correct classification: Picks method_class="system" and the right tags (with-llm, closed-source-api) so the system lands in the correct leaderboard entrant pools. - Packaging and verification: Adds the pip extra to pyproject.toml, runs registry tests with pytest, enforces ruff lint/format, and prepares the PR using the repository template. - Use Case: You want to benchmark the FLAML AutoML framework on TabArena. The Skill walks you through writing the wrapper, declaring two preset configs, tagging it as open-source, and verifying discovery via pytest tests/tabarena/systems/. ## Quick Start Ask the assistant to add FLAML as a system to TabArena with the flaml pip package and its documentation URL.
