AI hallucinates on legal questions. Lawbster delivers the facts.¶

Lawbster is the MCP server for German and European law. It connects Claude, ChatGPT, Copilot and any other MCP-capable AI assistant directly to ~11,000 statutes, federal court decisions and the entire EU legal corpus — updated daily from official sources, every match returned with an exact reference.

Instead of relying on a language model's training data — which is outdated, incomplete and not citable — LLMs talk to Lawbster over the open Model Context Protocol and answer with paragraph numbers, article numbers and the date of the last amendment.

Get connected in 2 minutes Tool reference Compliance

Headline benefits¶

Verified legal context

Every match comes from an official source: gesetze-im-internet.de, EUR-Lex, Rechtsprechung-im-Internet. No hallucinations, no fabricated paragraph numbers.
Updated daily

No training cutoff. Lawbster pulls from official sources nightly, parses changes and re-indexes affected provisions.
Hybrid search with AI reranking

Semantic plus keyword search, with an AI reranking step. MRR@10 = 0.676 on GerLeRB — transparent public benchmark.
Every answer is citable

§ 823 BGB, Art. 6 GDPR, BGH VI ZR 175/22 — exact paragraphs, correct numbers, cleanly separated between federal law, EU law and case law.
Hosted in Germany

Hetzner servers in Germany. No data transfer to third countries. Art. 28 GDPR DPA included.
Per-seat model, no lock-in

€19 / seat / month. 14-day free trial. Cancel any time. Add as many seats as you need — for people or standalone API keys.

What Lawbster covers¶

Source	Content	Updated
German federal law	BGB, StGB, HGB, AO, EStG, GewO, AGG, KSchG, ArbZG, MiLoG, …	Daily
EU law	GDPR, AI Act, MiCA, NIS2, regulations, directives	Daily
Case law	Decisions of BGH, BVerfG, BAG, BSG, BPatG, BFH	Daily

→ Full data sources overview

Eight tools — all batch-capable, all async¶

Every tool is designed for the LLM tool-use loop: compact responses on discovery tools (~4 k tokens), full text on detail tools, pagination via count / total / offset / has_more / next_offset / hint.

Tool	Purpose
`legal_search`	Hybrid search across all sources with filters
`legal_lookup`	Full text of a single norm by citation
`legal_lookup_batch`	Up to 20 lookups in one call
`legal_get_context`	Surrounding norms of a reference
`legal_find_citing_decisions`	Federal court decisions citing a specific norm
`legal_list_laws`	Available laws with abbreviation and currency
`legal_get_toc`	Table of contents of a law
`legal_get_stats`	Index and database statistics

Plus static resources (legal://rechtsrahmen, legal://filter_values, legal://eu_celex_registry), dynamic resource templates (legal://norm/{id}, legal://law/{source}/{abbr}, legal://eu/celex/{id}) and three prompts (legal_research, citation_resolve, compare_de_eu).

Quickstart — most common clients¶

Claude Desktop

Add an mcpServers entry to claude_desktop_config.json — done.

Walk-through
:simple-openai:{ .lg .middle } ChatGPT & Claude.ai (web)

OAuth flow — no code required.

Walk-through
Cursor

.cursor/mcp.json with bearer token.

Walk-through
Copilot Studio

Custom connector with full tool & resource support.

Walk-through

Who uses Lawbster¶

Legal-tech engineers plug Lawbster into compliance systems, document-review pipelines and custom chatbots via API key. Law firms connect their AI assistants directly to verified legal texts. In-house compliance teams use Lawbster for GDPR, AI Act and labour-law research without ChatGPT hallucinations.

Lawbster is not Harvey or Noxtua — we don't replace a full-stack legal AI product. Lawbster is the layer underneath: the citable legal-context provider. Bring your own LLM, bring your own workflow. Like Context7, but for law.

Trust & compliance¶

GDPR-compliant, Art. 28 DPA included, hosting at Hetzner Online GmbH (Germany).
No processing of personal data in normal operation — the index contains only public legal texts, tool-call payloads are not persisted.
API keys SHA-256-hashed in the database, plain text shown only once.
Stripe (payment processing) under the EU-US Data Privacy Framework.
Transparent benchmarks on GerLeRB, publicly reproducible.
Fair use: 60 requests/min per seat — no throttling under normal use.

→ Full compliance overview

Pricing¶

One plan, simple billing.

€19 / seat / month
Unlimited requests per seat (fair use 60 / min)
Full legal corpus included
Priority email support
14-day free trial — no credit card
Cancel any time

Start your free trial

A product by PLANIT // TECH GmbH¶

Made in Germany. Operated by PLANIT // TECH GmbH, specialised in privacy and legal-tech software.

Website GitHub support@planitprima.com