MichaelYang-lyxCommunity·44 Agent Skills Included

AIDABench

Benchmarking end-to-end data analytics on real-world documents

Evaluates how well models handle end-to-end data analytics over real spreadsheets, databases, and financial reports. Covers file generation, analytical question answering, and chart creation with multi-step, cross-file reasoning tasks. Includes sandboxed code execution, three dedicated evaluators, and ready-made skills for Excel cleaning, large-file processing, and image data extraction.

110|3

↗

npx skills add MichaelYang-lyx/AIDABench --all -g -y

All Skills in This Repository (44)

Pure Emerald Level Indicators

📦 In Repo

MichaelYang-lyx

sn-da-large-file-analysis

Converts large Excel files to Parquet using streaming reads and memory optimization.

sn-da-excel-workflow

Orchestrate multi-sheet Excel data workflows with Parquet optimization.

sn-da-image-caption

Caption charts, tables, UI screenshots, and diagrams into structured data.

excel-multi-sheet-threshold-analysis

Count rows across Excel sheets, deduplicate dimensions, and export summary and detail reports.

invalid-data-cleaning

Clean large Excel data with conditional Parquet conversion and regex text cleaning.

excel-smart-analysis-and-cleaning

Clean and reconcile multisheet Excel data with pandas and openpyxl.

text-normalization-and-large-file-processing

Normalize numeric prefixes and extract Chinese characters from Excel data.

numeric-format-normalization

Normalize numeric columns in Excel workbooks and verify totals against a Summary sheet.

outlier-detection-and-quality-assessment

Detect outliers and assess data quality in non-normal tabular datasets.

excel-bar-chart-visualization

Consolidate multi-sheet Excel data, clean merged cells, and generate bilingual bar charts.

stacked-chart-visualization

Convert percentage strings into numeric values and render stacked bar charts.

pie-chart-data-analysis

Analyze multi-sheet Excel and CSV data and generate a downloadable pie-chart report.

Community

Intermediate

Frequently Asked Questions

FAQPage Schema

How to install AIDABench?▼

Run `npx skills add MichaelYang-lyx/AIDABench --all -g -y` in your terminal to install all skills in this suite globally.

What does AIDABench evaluate?▼

It tests models on end-to-end data analytics tasks over real documents, covering file generation, question answering, and data visualization with up to 14 input files per task.

How does AIDABench score model outputs?▼

It runs models in a sandboxed Python environment and uses three dedicated evaluators to judge text answers, chart quality, and spreadsheet structure and content.

Can AIDABench handle very large Excel files?▼

Yes. Its skills include streaming reads, Parquet conversion, and memory optimization for files with 100k to over 1 million rows.

Does AIDABench work with different models?▼

Yes. It supports OpenAI-compatible APIs, Claude, and other providers through configurable agent runners and environment variables.