Human Rights Law as Data

Use with your AI assistant

Four of the tools can also be added to Claude, ChatGPT or another assistant that supports the Model Context Protocol (MCP). The connectors are hosted, so there is nothing to install: you add one address, and the assistant searches the same corpus as the website. The connector returns the text verbatim, with the UN symbol and paragraph (or the case name, application number and §), so the assistant can quote it instead of reconstructing it from memory and you can check every quotation.

However, the Free plan of Claude allows only one custom connector, so choose the corpus you need most. One address for all the corpora is planned.

HRC Voting, however, runs as an MCP server on your own computer (it works offline), and its repository explains how to install it.

Confidential text

The conversation with an assistant goes to the company that runs it (e.g., Anthropic for Claude, OpenAI for ChatGPT), so do not put confidential text into it, such as a draft submission or a client's file. The connectors themselves receive only what the assistant sends them in a search (the search terms or the citation to look up).

Claude (web and desktop app)

Open Customize and then Connectors, click Add and choose Add custom connector. Give it a name, paste the address, choose No sign in and click Add. In a new chat, switch the connector on in the tools menu. On Team and Enterprise plans an owner adds the connector first in Organization settings, and members then click Connect.

Claude help centre: custom connectors

ChatGPT

Open Plugins, select the plus button and then Add custom MCP server. Give it a name, paste the address as the MCP server URL, choose no authentication, confirm the risk warning and select Create as a plugin. In a new chat, type @ and choose it. Whether you can add a custom server depends on your plan and, on a work account, on the settings of your workspace.

OpenAI documentation: connecting a custom MCP server in ChatGPT

Claude Code

Run the command below in a terminal, and the connector is available in every project.

claude mcp add --transport http unhrdb https://150.254.115.204/unhrdb-mcp-rpc/mcp --scope user
claude mcp add --transport http echr https://150.254.115.204/echr-api/mcp --scope user
claude mcp add --transport http travaux https://150.254.115.204/travaux-api/mcp --scope user

Cursor and VS Code

The buttons next to each address open the editor and add the connector after you confirm.

Example prompts

  • What did the Human Rights Committee say in General Comment No. 37 about restrictions on peaceful assembly that must be “necessary in a democratic society”? Quote the paragraphs with their numbers.

  • Find the judgments of the European Court of Human Rights on bans of marches and pride parades, and quote the key paragraph with the case name, application number and paragraph.

  • What did the delegations in the Commission on Human Rights say between 1949 and 1952 about the words “in a democratic society” in the article that became Article 21 of the ICCPR? Cite each speech by UN symbol and page or paragraph.

  • Which recommendations on the right of peaceful assembly were made in the Universal Periodic Review in 2026? Quote five of them with the UN symbol, the State under review and the recommending State.

Check the answer

The assistant writes its own answer, and it may still misquote a paragraph or attach a quotation to the wrong one. For this reason, paste the answer into Check before you use it, which compares every quotation and citation with the source (Check in the UN Human Rights Database for UN documents, HUDOC Researcher Check for the judgments of the Court and Travaux Check for the drafting records). The first two keep your text in the browser and fetch only the identifiers of the cited documents, while Travaux Check reads the text on its server and keeps it nowhere.