Catalog
Tool catalog
Connect once and your AI can find and run any tool it needs. Below is what it can reach today: tools your agent pays for by the call.
Catalog results
7 tools in Machine Learning.
Hugging Face Dataset Configs and Splits
Lists the configurations and splits of one public Hugging Face dataset as the Dataset Viewer knows them: one entry per (config, split) pair…
$0.0026 per call
Hugging Face · Machine Learning
Hugging Face Dataset Details
Returns the Hugging Face Hub record of one dataset repo given its owner and name: downloads over the last 30 days, likes, dates, head…
$0.0026 per call
Hugging Face · Machine Learning
Hugging Face Dataset Info (Features, Splits, Sizes)
Returns the Dataset Viewer's info for one public Hugging Face dataset: for each configuration (or for the one `config` asked for) the…
$0.0026 per call
Hugging Face · Machine Learning
Hugging Face Dataset Rows (Paged, Cells Clipped)
Returns one page of rows from one split of a public Hugging Face dataset: the column list (`features`, with each column's type), the rows…
$0.0026 per call
Hugging Face · Machine Learning
Hugging Face Dataset Search
Searches Hugging Face Hub datasets and returns up to `limit` (1 to 25, default 10) of them ordered by `sort` (downloads, likes, createdAt,…
$0.0026 per call
Hugging Face · Machine Learning
Hugging Face Model Details
Returns the Hugging Face Hub record of one model repo given its owner and name: downloads over the last 30 days, likes, task…
$0.0026 per call
Hugging Face · Machine Learning
Hugging Face Model Search
Searches Hugging Face Hub models and returns up to `limit` (1 to 25, default 10) of them ordered by `sort` (downloads, likes, createdAt,…
$0.0026 per call
Hugging Face · Machine Learning
Every tool has an explicit pricing model and a usage guide written for it and checked against its schema.
You pay the price shown. If a call turns out to cost more than its quote, you pay the actual price, capped at twice the quote. Run omnial_execute with dry_run: true for an exact quote.
Search here runs the same omnial_search call your agent makes over MCP: ranked matches, each showing the keyword or field it matched on. It always answers with what it found; an empty result means the catalog carries none of those terms, and the full list is right below.