Litigation MCP Server

Overview

The server exposes 4 tools over the Clarivate Caselaw Database. Search finds and analyzes cases, one tool returns full case details, and two help you discover valid field values.

Add the server URL above to any MCP-compatible client as a remote server. Your assistant then chooses the right tool from the descriptions below. Some clients show tool names with an ip-one-litigation___ prefix.

Data is organized as cases and documents. A case is identified by a GUID, and several court documents can belong to the same case. Returned records are always distinct.

Workflows

Find and review cases

1. 1. Discover

   Look up the exact values a filter needs, such as court names or party names, when a wildcard search misses.

   search_field_valuesdescribe_fields
2. 2. Search

   Filter cases by domain, party, court, dates and more. The result is a list of case GUIDs and a total count.

   search_cases
3. 3. Review

   Fetch the full details of up to 50 cases at a time.

   get_case_details

{
  "tool": "search_cases",
  "arguments": {
    "filters": {
      "cases_filter": [
        { "field": { "root": "CASE_DOMAIN" }, "op": "EQ", "value": "PATENT" },
        { "field": { "root": "PARTY_OPTIMIZED_NAME" }, "op": "LIKE", "value": "%ACME%" },
        { "field": { "root": "FIRST_ACTION_DATE" }, "op": "GE", "value": "2020-01-01" }
      ]
    },
    "limit": 10
  }
}

Analyze with grouping and aggregation

Use group_by with aggregate to count or summarize cases instead of listing them. Only the grouped and aggregated fields are returned, and order_by can sort on an aggregated field such as GUID_COUNT.

{
  "tool": "search_cases",
  "arguments": {
    "filters": {
      "cases_filter": [
        { "field": { "root": "CASE_DOMAIN" }, "op": "EQ", "value": "TRADEMARK" }
      ]
    },
    "group_by": [ { "root": "DOCKET_COURT_COUNTRY" } ],
    "aggregate": [ { "func": "COUNT", "field": { "root": "GUID" } } ],
    "order_by": { "field": { "root": "GUID_COUNT" }, "direction": "DESC" },
    "limit": 10
  }
}

Read the cases you found

{
  "tool": "get_case_details",
  "arguments": {
    "guids": [ "<case GUID from search_cases>" ]
  }
}

Tool reference

Fields are named in uppercase, for example CASE_DOMAIN, and are passed as objects: {"root": "CASE_DOMAIN"}. See the field reference for every field.

Search

Case details

Discovery

Field reference

These fields can be used in filters, group_by and aggregate. String values are stored in uppercase.

Case

Parties

Court

Documents

Patents

Trademarks

Designs

Copyright

Query guide

• Use uppercase or LIKE for text String fields are indexed in uppercase, so use uppercase values or the LIKE operator. LIKE supports prefix (APPLE%), suffix (%SAMSUNG) and contains (%GOOGLE%) matches.
• Match names loosely For party names, prefer LIKE unless you need an exact match, and check that the results are the parties you meant. If LIKE finds nothing, use search_field_values to get the exact values and retry.
• Enumerated fields need exact values Fields with a fixed list of values, such as PLAINTIFF or DEFENDANT, require an exact match.
• Ranges and classifications Date and numeric fields support GE and LE for ranges. Classification fields use the official Nice, Locarno and IPC codes, and Nice and Locarno codes vary in format, so use LIKE to cover the variants.
• Scope by industry with IPC codes For industry or product questions where no company is named, such as hearing aids or automotive patents, filter on IPC_CLASS_CODE first.
• Grouping, ordering and paging Aggregation requires group_by. order_by must use a field from group_by or aggregate, and null values sort to the top when ordering descending, so filter them out with a minimum value. The response always includes the total record count, so you don’t need a grouped count for that. limit is capped at 50 per call, and get_case_details takes up to 50 GUIDs.
• Keep requests focused Request no more than 10 fields at a time. When browsing cases without other instructions, group by EARLIEST_EVENT_DATE and sort descending.