get_case
ChatGPTGet detailed information about a single court case by its Trellis docket ID. Returns full case details including entities by role (parties, counsel, experts), docket events, and classification. Events are paginated (default page size 25, max 100) and sorted by date descending. Use the events_* parameters to filter by date range, entry type, or PDF availability. total_events is the unfiltered total on the case; filtered_events_count is the post-filter total. To find cases and their docket IDs, use the search_cases tool. Available output_fields: access, no_results, docket_id, case_number, case_name, state, county, court_name, filing_courthouse, judge, judge_id, filing_date, case_type, status, classification, entities, events, total_events, filtered_events_count, events_page, events_page_size, file_count, verdict, causes_of_action, complaint_summary, relief_sought, facts, date_of_losses, url
get_case_aggregations
ChatGPTCount and rank cases by any dimension WITHOUT retrieving case results. Returns the total case_count plus top-N bucket counts for each requested group_by dimension, computed over every case matching the filters. Accepts the same filters as search_cases. Prefer this over search_cases whenever the question is analytical — 'how many', 'which X most often', 'top judges/counties', distributions, or filing trends over time — it answers in a single call instead of paginating through case results. Available output_fields: access, no_results, case_count, aggregations, filing_trend
get_case_complaints
ChatGPTGet all the complaints for a case with the primary being marked as the main one. Automatically finds the original complaint document. Only the main complaint is returned by default, but you can include other complaint filings like amended complaints and supplemental complaints by setting only_main_complaint to False.Document text content is NOT included by default — to retrieve it, include 'content' in output_fields. Useful for extracting case details like property addresses, causes of action, and factual allegations. When pdf_download_url is present, show it as a markdown 'Download PDF' link next to the document. Downloading requires the user to be logged in to trellis.law in their browser and have website download access. Available output_fields: total_results, page, page_size, aggregations, access, no_results, pdf_id, pdf_download_url, docket_id, case_number, case_name, state, county, court_name, judge, filing_date, event_description, practice_area, matter_type, motion_type, outcome_type, entry_type, filer_type, page_count, duration_days, content, plaintiffs, defendants, experts, causes_of_action, case_summary, is_class_action, is_bad_faith_insurance, docket_disposition, url, is_main_complaint
get_county_coverage
ChatGPTGet detailed coverage info for ONE county. Returns: county metadata (population, seat, established year, etymology, area, form of government, board of supervisor, description), contact info (website, phone numbers, administration & court addresses), the list of courthouses (name/address/website), practice areas with at least one docket on file, and a has_documents flag. Use this when the user asks about a specific county. For state-wide info use get_state_coverage.
get_document
ChatGPTGet a court filing by its PDF document ID. Returns document metadata and classification. Document text content is NOT included by default — to retrieve it, include 'content' in output_fields. Use max_content_length to control how much text is returned. When pdf_download_url is present, show it as a markdown 'Download PDF' link next to the document. Downloading requires the user to be logged in to trellis.law in their browser and have website download access. Available output_fields: access, no_results, pdf_id, pdf_download_url, docket_id, case_number, case_name, state, county, court_name, judge, filing_date, event_description, practice_area, matter_type, motion_type, outcome_type, entry_type, filer_type, page_count, duration_days, content, plaintiffs, defendants, experts, causes_of_action, case_summary, is_class_action, is_bad_faith_insurance, docket_disposition, url, is_main_complaint
get_judge_analytics
ChatGPTGet analytics for a judge by their ID, available from search_judges tool. Not all judges have an analytics dashboard available. By default returns only case_analysis (caseload, average case length, motion grant rate, plaintiff/defendant grant rates, practice-area mix). Pass output_fields to opt into additional sections: motion_analysis (per-motion-type grant/deny/partial rates with plaintiff/defendant split and court+state volume context), case_milestones (cycle time, time to trial, time to first dismissal, time to CMC), or matter_type_outcomes (disposition mix). Per-practice-area detail is opt-in via dot paths: 'case_milestones.by_practice_area' or 'matter_type_outcomes.by_practice_area'. Nested matter-type detail under each practice area requires the deeper opt-in: 'case_milestones.by_practice_area.matter_types' or 'matter_type_outcomes.by_practice_area.matter_types'. All metrics include county and state benchmarks where available. Available output_fields: access, no_results, case_analysis, motion_analysis, case_milestones, matter_type_outcomes, last_updated, judge_url
get_judge_profile
ChatGPTGet a detailed judge profile by judge ID. Returns identity (fullname, parsed name, court, county, state), biography, appointing authority, clerk phone, bar admissions, news articles about the judge, courtroom rule documents, and career history. Use search_judges first to find the judge_id. For judge analytics (case load, motion grant rates, milestones) use get_judge_analytics. Available output_fields: access, no_results, judge_id, fullname, parsed_name, state, county_name, court_name, bio, appointed_by, clerk_phone, bar_certifications, articles_about, courtroom_rules, career_history, judge_url, status
get_jurisdiction_analytics
ChatGPTGet cycle-time analytics for a jurisdiction: how long cases take, in days, for a given state / county / practice area / matter type. Returns the average days from filing to disposition (avg_cycle_time) and to first trial (time_to_trial), plus the docket sample_size behind those numbers. Use this for questions like 'average time to trial for breach-of-contract in Miami-Dade'. Numbers are computed over resolved cases only. Available output_fields: access, no_results, state, county, practice_area, matter_type, avg_cycle_time, time_to_trial, sample_size
get_matter_types
ChatGPTGet valid matter types, optionally filtered by practice area. Matter types are sub-classifications within a practice area (e.g. 'Automobile' and 'Premises Liability' are matter types under 'Torts'). Use this before passing a matter_type filter to search_cases.
get_ruling
ChatGPTGet a single court ruling by its ruling ID. Returns full ruling text and metadata. Use search_california_rulings to find ruling IDs. Available output_fields: access, no_results, ruling_id, case_number, case_name, judge, state, county, date, ruling_type, ruling_subtype, ruling_text, docket_id, case_type, department, status, court_name, parties, ruling_outcome_summary, motion, practice_area, matter_type, is_class_action, display_county, url
get_server_instructions
ChatGPTReturn MCP server instructions (if any). Always call first.
get_specialties
ChatGPTGet valid specialties, optionally filtered by specialty group. Specialties are sub-classifications within a specialty group (e.g. 'Neurology' and 'Pathology' are specialties under 'Medical'). Use this before passing a specialty_group filter to search_experts.
get_state_coverage
ChatGPTGet Trellis coverage for a single state. Returns the state-level coverage flags (civil, family, probate cases; documents; verdicts; judge bios and analytics; state law; motions and issues; tentative rulings) and the list of covered counties (name, courthouse count, and whether Trellis has indexed documents for the county). To compare multiple states in a matrix, use show_coverage_table. For deep details about a single county, use get_county_coverage.
get_support_article
ChatGPTFetch the full body of a Trellis support article by its support.trellis.law URL. Call this after search_support_articles so you can read the article and answer the user's question with specifics.
get_usage
ChatGPTReturns the number of MCP views consumed and remaining for the current quota period. Use this to check how much of your MCP view quota has been used before running large queries. Does not count toward your MCP views.
search_california_rulings
ChatGPTSearch California tentative rulings — preliminary written decisions that California state trial court judges post before a hearing, indicating how they intend to rule. Tentative rulings become final if uncontested and are strong predictors of the eventual outcome. They contain the judge's actual written reasoning. For all other judicial rulings/orders (including from other states), use search_documents. Available output_fields: total_results, page, page_size, aggregations, ruling_id, case_number, case_name, judge, state, county, date, ruling_type, ruling_subtype, ruling_text_snippet, docket_id, case_type, department, status, court_name, parties, ruling_outcome_summary, motion, practice_area, matter_type, is_class_action, display_county, url
search_cases
ChatGPTSearch state court cases/dockets. Returns case metadata with aggregated facet counts for filtering. For further details about a case, use the get_case tool. Available output_fields: total_results, page, page_size, aggregations, docket_id, case_number, case_name, state, county, court_name, judge, filing_date, practice_area, matter_type, is_bad_faith_insurance, docket_disposition, status, case_type, entity_summaries, entity_counts, all_entities_count, url, verdict, is_class_action, disposition, disposition_date, filing_courthouse, case_last_updated, duration_days, causes_of_action, complaint_short_summary
search_documents
ChatGPTSearch court filings and PDF documents. Searches full document text content including complaints, motions, and orders. Returns document metadata with content snippets and aggregated facets. Use the get_document tool to retrieve further details about the document including the full document text content. When pdf_download_url is present, show it as a markdown 'Download PDF' link next to the document. Downloading requires the user to be logged in to trellis.law in their browser and have website download access. Available output_fields: total_results, page, page_size, aggregations, pdf_id, pdf_download_url, docket_id, case_number, case_name, state, county, court_name, judge, filing_date, event_title, event_description, content_snippet, practice_area, matter_type, motion_type, outcome_type, entry_type, filer_type, entity_summaries, page_count, status, duration_days, url
search_experts
ChatGPTSearch expert witness profiles by name, state, or specialty. Returns expert profiles with specialties, pdf counts, docket counts, and states where the expert has appeared. To get cases or documents associated with the experts, use the search_cases or search_documents tools.Does NOT return cases or documents associated with the experts. Does NOT contain all experts in the system. Not all documents have had experts extracted and indexed. Available output_fields: total_results, page, page_size, total_pages, expert_id, full_name, title, first_name, middle_name, last_name, suffix, nickname, specialties, first_seen_date, last_seen_date, count_pdfs, count_dockets, states
search_judges
ChatGPTSearch for judges by name, state, or county. Returns judge profiles with judge_id values that can be used with get_judge_profile or get_judge_analytics tools. Available output_fields: total_results, page, page_size, judge_id, fullname, state, county_name, court_name, judge_url, has_analytics, status
search_support_articles
ChatGPTSearch the Trellis customer support knowledge base (https://support.trellis.law) for articles relevant to the user's question. Use this when the user asks how to do something in Trellis, reports a feature/billing/account issue, or wants product documentation. Returns a list of articles with title, URL, and a long snippet. Follow up with get_support_article to read the full body before answering.
