MCP server and CLI for the Deutsches Zeitungsportal (Deutsche Digitale Bibliothek) - search and retrieve German newspaper pages
Four well-named tools with clear, domain-specific descriptions (194 - 392 chars, within baseline). All tools have complete input schemas with typed parameters and descriptions. Output schemas are documented in docstrings. Naming follows verb_noun pattern (search_, list_, get_, download_). No critical security issues detected. Main gaps: no error handling guidance, no enum constraints for 'field' parameter in list_ddb_facet_values, and no explicit output field documentation for structured responses.
Download the OCR text of a page, or of every page of an issue.
Find where a query appears inside one page or one whole issue.
List the values a search filter can take, with page counts. Use this before filtering, and to characterise a result set. `place` and `provider` are matched as whole strings, so place='Halle' returns zero pages while place='Halle (Saale)' returns 1.1M - this is how to find which form the index holds. Given a query, the counts describe that result set instead of the whole corpus, which answers "where did this term appear, and in which newspapers".
Search German newspaper pages in the Deutsches Zeitungsportal. Covers ~33.8M digitised pages, densest between 1850 and 1949. Results are individual newspaper *pages*, and each carries highlighted snippets showing where the query matched, with matched terms in {braces} - so a single search is usually enough to judge a hit without downloading anything.
list_ddb_facet_values 'field' parameter lacks enum constraint. Description lists valid values ('place', 'provider', 'language', 'title') but no formal enum in schema, inviting hallucinated values.
No error handling guidance. Tools return dict/str but lack documentation of error cases, recovery steps, or what to do if API fails (e.g., 'If search returns 0 results, try broadening the query or removing year filters').
Output schemas not formally documented. Docstrings describe return values in prose ('total_results', 'total_pages', 'matching pages') but no structured JSON Schema for response fields, forcing LLMs to infer structure.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 63 | 2026-07-28+ | v2 |
download_ddb_text returns a file path (string) rather than file contents or a structured object with path + metadata. Docstring says 'Path to the cached text file' but LLMs expect content or a reference they can act on.
No pagination guidance for search_ddb results. While 'page' and 'rows' parameters exist, docstring does not explain max result limits, when pagination is needed, or how to iterate through large result sets efficiently.