What it does
The server wraps NLP tools from LINDAT/ÚFAL MFF UK and exposes them through MCP. It offers named entity recognition for Czech and 34 other languages, with automatic language detection, and morphological analysis for 35 languages. Its anonymization tool uses regex patterns for more than 80 kinds of personal data, covering Czech identifiers and international formats such as IBAN, EU VAT numbers, US SSNs, several national ID numbers, crypto addresses and API tokens. It also provides Czech readability checks, Czech spell correction and translation among 8 languages. An optional placeholder mode gives consistent labels such as OSOBA1, and an offline local mode is available.
It suits workflows where personal data must be removed before text goes to an AI tool, such as legal files, medical records, banking statements, insurance claims, notarial documents and research corpora. The publisher reports testing it on a 12.7 KB multi-sector document, where it caught 94 of 94 unique PII items in one call.
After installation from PyPI, the server is registered in an MCP client, which then gets six tools. The publisher documents setup for Claude Code and for older versions of Claude Desktop, through a configuration file.
Limitations and requirements
Non-commercial use only. The NameTag and UDPipe models are licensed under CC BY-NC-SA, and the LINDAT API is free only for academic and personal use. Commercial deployment requires contacting the tool authors and the ÚFAL contact address given by the publisher. This is an unofficial community project, not operated or approved by ÚFAL MFF UK. The newer Claude Desktop supports only remote MCP servers over HTTP, so local stdio servers such as this one are not supported there.
Safety
Our team has not reviewed this resource. Only automated rules checked it; resources with high-risk signals are not published. These are signals, not a guarantee that installing it is safe. Try it in a sandbox project first.
- Runs code or changes an assistant's behaviour
Sources
- Official MCP Registry first seen 13 Nov 2026 · last seen 13 Nov 2026 · its page on the source
Czech PII Anonymizer & NLP
Anonymize PII & redact text (GDPR): Czech + 35-lang NER, morphology, translation, spellcheck
v0.10.2As shown on registry.modelcontextprotocol.io when we last read it