The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the KnowJudges listing page.
Add source-linked U.S. judge and court research directly to AI assistants and agents through the Model Context Protocol.
Website: https://knowjudges.com
MCP documentation: https://knowjudges.com/mcp
Remote MCP endpoint: https://mcp.knowjudges.com/mcp
The KnowJudges MCP server exposes six focused research tools over Streamable HTTP.
Compatible MCP clients can discover the available tools, validate inputs, and retrieve structured, source-linked KnowJudges results without building a custom API integration.
Public access currently requires no authentication token.
Use this remote MCP endpoint:
Transport: Streamable HTTP Authentication: None required for public access
For clients that support an MCP server configuration object:
Configuration field names vary between MCP clients. If your client asks only for a server URL, enter:
The official KnowJudges connection instructions use this same remote endpoint and transport.
find_judgesResolve a likely judge and canonical KnowJudges slug.
Use this when you know some combination of a judge's name, court, or jurisdiction and need to identify the correct KnowJudges profile.
get_judge_profileRetrieve a public judge profile containing source-linked, scoped findings.
get_judge_biographyRetrieve sourced public background information about a judge without behavioral inference.
get_courtroom_presentation_rulesResearch courtroom presentation requirements and guidance, including topics such as:
find_ai_abuse_casesFind up to 10 court cases involving AI-generated legal errors.
Results can include:
get_recent_rule_changesRetrieve changes recorded since KnowJudges change tracking began.
These are the six tools currently exposed by the public MCP service.
After connecting the MCP server, try:
Find the five most recent New York court cases involving fabricated AI-generated legal material. Include the judge, court response, and verified official decision URL when available.
A provenance-aware response should preserve the underlying source information and avoid making claims beyond what the published records support.
The public KnowJudges REST API and MCP server share the same allowance for each public client or network:
| Window | Allowance |
|---|---|
| Hourly | 20 requests |
| Daily | 100 requests |
Requests exceeding either limit return HTTP 429.
MCP initialization, tool discovery, and individual tool calls are separate HTTP requests and each counts toward this shared allowance.
For datasets, recurring feeds, batch research, higher-volume usage, or private integrations, see:
KnowJudges is designed around source-linked research.
Applications and agents should preserve source information returned by MCP tools, including official decision links and other provenance metadata where available.
In particular:
A typical agent workflow might be:
For legal-AI research:
KnowJudges also provides a public REST API for applications that do not use MCP.
API documentation:
MCP is generally the simpler interface for MCP-compatible AI assistants and agents, while the REST API provides direct HTTP access for conventional applications.
https://mcp.knowjudges.com/mcpKnowJudges provides free, source-linked research on U.S. judges and courts through public judicial profiles, court information, research tools, API access, and MCP integrations.
The MCP server is intended to make that information directly usable by AI assistants, research agents, and other MCP-compatible software while preserving the provenance of the underlying research.
For current terms, data-use information, licensing information, and corrections policies, refer to the relevant pages on https://knowjudges.com.