search_laws

Search normalized legal texts with optional law-code filters.

Parameters

Name	Type	Required	Description
query	string	yes	Search query string (keywords, phrases).
codes	string[]	no	Limit search to specific law codes (e.g. ["bgb", "dsgvo"]).

Returns

{
  results: Array<{
    code: string;
    norm: string;
    heading: string;
    score: number;      // relevance score (higher is better)
    text_preview: string;  // matching excerpt, ~200 characters
  }>;
  total: number;
}

Example

result = mcp_client.call_tool("search_laws", {"query": "Treu und Glauben"})

{
  "results": [
    {
      "code": "bgb",
      "norm": "par:242",
      "heading": "Leistung nach Treu und Glauben",
      "score": 0.95,
      "text_preview": "Der Schuldner ist verpflichtet, die Leistung so zu bewirken, wie Treu und Glauben..."
    }
  ],
  "total": 1
}

Scoped to specific laws:

result = mcp_client.call_tool("search_laws", {
    "query": "personenbezogene Daten",
    "codes": ["dsgvo"]
})

Notes

• Search uses fuzzy matching via rapidfuzz on normalized text.
• Results are ranked by relevance score and capped at a bounded result set.
• Passing codes significantly narrows the search space and improves result relevance.
• The search index is built at corpus load time; runtime queries are fast.

Related

• get_norm — fetch full text when you have the norm path.
• resolve_citation — look up a specific citation.
• list_laws — discover available law codes to use as filters.

Source

mcp/server.py — @app.tool() definition for search_laws.