Legal sources with official primary sources

Printed ·

Skip to main content

Verified Legal Source Infrastructure

We keep no copies of decisions or statutes. When you or a machine requests a document, it is retrieved directly from the publisher's own store, the Swedish courts, the Riksdag, EUR-Lex or the US court, and passed on with an integrity proof. What we hold is the structure: identifiers, statute links, relations between decisions, read dates and a citation line.

The original always comes from the publisher

Traditional legal databases keep their own copies that must be kept in step with the source. Here the publisher is always the source. You receive the version the publisher holds at retrieval, never an older transcript.

An integrity proof for every retrieval

Every retrieval carries a SHA-256 checksum of exactly what the publisher delivered, the retrieval time and the address it came from. A lawyer or an AI agent can later show what was retrieved, when, and that it was not altered on the way.

Source authority as data

Every document states the publisher, the publisher type (official, mirror or derived) and whether the publisher is the source of record, so sources can be ranked.

Short cache, no permanent storage

To spare publishers and answer quickly, a document is cached for at most one hour, statutes for 15 minutes. After that it is retrieved again.

The same basis as the decision agents

The CAGI agents and NovaCopilot rest on the same addresses. Every source in a decision bundle can be retrieved in original with the same integrity proof.

Example integrity proof

{
  "source_authority": {
    "publisher": "Sveriges riksdag",
    "publisher_type": "official",
    "source_of_record": true
  },
  "integrity": {
    "publisher_hash": "sha256:…",
    "retrieved_at": "2026-09-25T18:02:11Z",
    "publisher_url": "https://data.riksdagen.se/dokument/sfs-1962-700.html",
    "verified": true
  }
}

For machines