Pandex launches API to bring European law to AI agents
Pandex has released a specialized API and MCP server providing access to 2.8 million official European legal documents for AI-driven compliance.
Source: Hacker News — front pagePandex has launched a specialized API and Model Context Protocol (MCP) server designed to provide AI agents and software with direct access to official European legislation and high-court decisions. This development aims to solve the problem of legal hallucination by grounding AI responses in verified, primary-source legal texts.
What is the Pandex legal API?
According to Pandex, the platform serves as a single point of access for the laws and highest-court decisions of the European Union and seven specific European countries. The service provides a massive corpus of data, including 2.8 million official documents, 421,000 acts and codes, and 2.4 million court decisions.
Rather than relying on web scrapers that must be maintained for every individual court, Pandex offers a unified REST API and an MCP server. This allows developers to integrate high-fidelity legal data into existing AI workflows, such as those used in coding assistants or specialized legal research tools. The platform supports semantic, keyword, and hybrid searches across different languages, enabling a user to ask a question in English and receive a specific passage from a French ruling.
Which jurisdictions and documents are covered?
As reported by Pandex, the current coverage is extensive but varies in completeness depending on the institution. The platform holds complete records for several key entities, including the Publications Office of the EU and the Court of Justice of the EU.
In France, the service includes complete data from the Parlement, Gouvernement, Conseil d'État, and others, though the Cour de cassation is noted as partial due to the way certain decisions are published. In Germany, the platform has complete coverage for the Bundesgerichtshof and several other federal courts, though the Bundesverfassungsgericht is currently partial. The Netherlands, United Kingdom, Austria, Switzerland, Luxembourg, and the Council of Europe are also included in the current rollout.
According to the provider, the platform also tracks "point-in-time" law. This means users can access previous versions of specific articles and see exactly when changes took effect, providing a full citation graph for compliance and historical research.
How does this affect AI agents and developers?
For developers building agentic workflows, the introduction of an MCP server is significant. MCP allows AI models to interact with external data sources more seamlessly. Pandex is designed to work with popular AI tools and environments, including Claude, Cursor, GitHub Copilot, Windsurf, Zed, and Gemini.
By using this API, an agent can move beyond general knowledge and perform "exact lookups" within official texts. For instance, instead of an AI guessing the rules regarding employment status, it can query the specific case law from the Cour de cassation to find the exact criteria for a subordinate relationship. This level of grounding is intended for software that requires high levels of accuracy for compliance and monitoring tasks.
What is the pricing model for Pandex?
According to the Pandex website, the service follows a tiered pricing structure. There is a free tier available during the beta period, which allows for 1,000 requests per day via the REST API and MCP server.
For commercial products, the usage-based plan costs €1 per 1,000 requests. There is also a "Feed" option starting from €1,000 per month, which provides the entire corpus as files for companies that wish to index the data within their own internal systems. Additionally, Pandex is offering a "founding customer" program for the first ten companies, priced at €300 per month for a fixed two-year period, which includes access to the API without request limits up to 1 million per month.
How can researchers use this data?
Researchers can utilize the platform to perform deep legal analysis through the linked citation graph. Because the data is pulled straight from the institutions that publish it, it offers a level of verification that is often missing in general-purpose large language models. The ability to search through 112 million searchable passages and 37 million linked citations allows for complex queries regarding how specific laws have evolved or how different courts have interpreted similar statutes over time.
For those working with complex legal papers or multi-jurisdictional research, this structured data provides a foundation for more reliable automated reasoning. While general AI models are prone to misinterpreting nuances in legal language, an agent equipped with the Pandex API can verify its logic against the actual text of the law.
In the context of academic or professional research, a user might explore these complex legal topics using Reader in Grafy. While Pandex provides the raw API for developers to build specialized tools, Grafy Reader allows researchers to upload their own legal PDFs and papers to perform deep, cited searches and questioning within their own collected documents.
Frequently asked questions
What jurisdictions does Pandex currently cover?
Pandex provides access to the legal documents of the European Union and seven specific countries: France, Germany, the Netherlands, the United Kingdom, Austria, Switzerland, and Luxembourg. It also includes data from the Council of Europe. The coverage includes both national legislation and the decisions of the highest courts in these regions.
How does the Pandex API support AI agents?
The Pandex API supports AI agents by providing a REST API and an MCP server. This allows models to perform exact lookups and semantic searches across millions of official legal documents. By connecting to these sources, agents can ground their answers in verified law and provide precise citations rather than relying on general training data.
Does Pandex offer a free version for testing?
Yes, Pandex offers a free tier while the product is in beta. This tier allows users to make up to 1,000 requests per day using the REST API and MCP server. This version provides access to all the data held by the platform, making it suitable for small agents and initial testing.
Source: Hacker News — front page — Pandex – Official European Law for AI Agents, over MCP


