ElevenFlo

Use when you need valid values for a document_type_tags filter. Inputs: optional query narrows category names or tags. Returns filing categories, each with its key, label and the document type tags it includes.

list_document_types

Use when the user asks which monitors are saved in their account. Inputs: optional offset and limit for pagination. Returns the user's monitors, oldest first, with monitor_id, filters and topics, total count and has_more.

list_monitors

Use when you need to know which structured datasets are available. Inputs: optional limit and cursor for pagination. Returns each dataset's slug (the dataset value other dataset tools take), title, description, what one row represents, how current the data is, record count, caveats and the tools that query it.

list_structured_datasets

Use when you already have passage IDs for one document, from search_document_chunks or extract_document_passages, and need their exact text or the passages next to them. Inputs: document_id and chunk_ids identify zero-based passage IDs in that document. Returns up to 8 passages of extracted text with page numbers where available (audio has none) and the filing_url to cite. Passage IDs are zero-based; when some requested IDs are out of range, the result lists missing_chunk_ids and total_chunks; if none exist, a chunk_not_found error gives the valid range. Text is extracted by OCR and can differ from the wording of the original PDF.

read_document_chunks

Use when the user asks to create or edit a saved monitor. Inputs: filters, optional monitor_id for an edit, name and topics. A case watch sets filters.case_id alone; a market monitor requires Pro and combines filters with AND. Editing replaces the saved filters and supplied topics; omitted topics remain unchanged. Edits can stop feed and email updates, make pending emails ineligible and invalidate list_case_updates cursors with monitor_changed. Returns the saved monitor, whether it was created, matching publication count for the last 28 days and counts excluded for unknown facts.

save_monitor

Use when you want a quick preview of where keywords appear across up to 25 filings. Inputs: document_ids (up to 25) and a non-empty keyword query. Returns for each filing its filing_url and the matching passage IDs with short previews and page numbers, plus missing_document_ids for any requested document that was not found. Every keyword must appear in a passage.

search_document_chunks

Use when you want to find filings in one case by what their summaries say. Inputs: case_id, keyword query and optional limit (up to 25). Returns up to 25 filings with the summary, filing_url and text_completeness. Summaries are generated automatically and can omit details, and only filings that have a summary are searched.

search_document_summaries

Use when you want to find court filings that mention specific terms, across all cases or within one case_id. Inputs: keyword query; optional case_id, document_type_tags and limit (up to 25). Returns up to 25 matching filings, newest first, with document_id, filing_url, case, document name, docket number, filing date, document type tags and a matching excerpt. Every keyword must appear somewhere in a filing. Quoted text is an exact phrase; uppercase OR, NOT and parentheses combine terms. A malformed query returns invalid_query. document_type_tags values come from list_document_types; type classification can be incomplete, so a type filter can leave out relevant filings.

search_filings

Use when you need individual records from a structured dataset, such as specific cases matching filters. Inputs: dataset; optional where filters, select columns, sort, facets, cursor and page_size. Returns matching rows with the selected columns, optional facet counts and a cursor for the next page. Filters, columns, sorts and facets are the ones listed by describe_structured_dataset.

search_structured_data

Use when you want to find court hearing transcripts that mention specific terms, across all cases or within one case_id. Inputs: keyword query; optional case_id and limit (up to 25). Returns up to 25 transcripts with document_id, filing_url, name, date and match counts, without transcript text. Every keyword must appear in a transcript.

search_transcripts

Use when you need the exact stored values of a dataset field, such as court or industry names. Inputs: dataset and field; optional query, where filters and limit. Returns matching values with labels and variant counts, optionally narrowed by a text query and filters.

suggest_structured_values