german-archives-mcp

An MCP server for German archives' finding aids and the vital registers German state archives have put online: search what German archives, libraries and museums have described in the Deutsche Digitale Bibliothek and its archive portal Archivportal-D, read one entry with its place in the archive's fonds, find the civil or church register of a place for a year at the Landesarchiv Baden-Württemberg, the Landesarchiv Nordrhein-Westfalen, the Landesarchivverwaltung Rheinland-Pfalz or the Hessisches Landesarchiv, and download a page of it to read.

The DDB is Germany's counterpart of DPLA, and for archives its counterpart of ArchiveGrid as well: archives deliver their finding aids to it, item by item, with signatures (call numbers), dates and the fonds each item belongs to. Some deliver images. Its version 2 API answers without a key.

The server covers every fonds of vital registers the Landesarchiv Baden-Württemberg has online (the Staatsarchiv Wertheim's civil-register duplicates, from 1870, are not yet in its online finding aids):

The Landesarchiv Nordrhein-Westfalen has put its civil registers (Zivilstands- and Personenstandsregister) online: the Rhineland's from 1798, and Westphalia's and Lippe's with some of their church-book duplicates, as far as the protection periods allow (births 110 years, marriages 80, deaths 30). The server searches them through the Archive in NRW portal and reads their pages from the Landesarchiv's image server.

The Landesarchivverwaltung Rheinland-Pfalz's Personenstandsarchiv, part of the Landeshauptarchiv Koblenz, has put the civil registers of the state's north online in its virtual reading room APERTUS: Best. 657,001 onward, one fonds per Standesamt or Bürgermeisterei, mostly the former government districts of Koblenz and Trier, from 1798. About 59,000 birth, marriage and death registers have images there: by the archive's count (March 2026) some 32,000 whole registers, digitised district by district up to 1880 or 1895, and 28,000 registers of which only the name index is digitised. APERTUS delivers a register only as one PDF of the whole volume, so the server downloads a volume once, keeps it on disk for a few days, and takes each page asked for out of it.

The Hessisches Landesarchiv's Personenstandsarchiv keeps the secondary registers of every Standesamt in today's Hesse, from 1874 or 1876, in fonds HStAM 900 to 926, and has put them online to 1900 (births), 1930 (marriages) and 1965 (deaths). The server finds them in the catalogue of the Landesgeschichtliches Informationssystem Hessen (LAGIS) and reads their pages from Arcinsys Hessen.

It works the way a careful genealogist does. A finding-aid entry says where a record is, not what it says: cite the archive and its signature, not the DDB. The page image is the evidence: read it before citing it, and cite the archive, the signature and the image number with the archive's permalink. A zero is not a negative: most German registers are not online, and many archives deliver nothing to the DDB.

Nothing here writes anywhere, and nothing here keeps a family tree. It sits beside dpla-catalog-mcp for American collections and snac-archives-mcp for finding which archive holds a family's papers.

This is an independent project. It is not affiliated with, endorsed by, or supported by the Deutsche Digitale Bibliothek, the Landesarchiv Baden-Württemberg, the Landesarchiv Nordrhein-Westfalen, the Landesarchivverwaltung Rheinland-Pfalz, the Hessisches Landesarchiv, LAGIS, or any archive whose finding aids it reads.

Tools

The server publishes eleven tools. All but the four image tools are read-only; each of those writes one new file and never overwrites one (rlp_image also keeps the volume PDFs it downloads in the cache folder).

The Deutsche Digitale Bibliothek and Archivportal-D

The registers at the Landesarchiv Baden-Württemberg

The civil registers at the Landesarchiv Nordrhein-Westfalen

The civil registers in Rheinland-Pfalz (APERTUS)

The civil registers in Hesse (LAGIS and Arcinsys)

Housekeeping

Setup

You need Python 3.11 or later and uv. There is no key to request.

Without cloning. uvx fetches it from PyPI and runs it in one step:

uvx german-archives-mcp

From a clone, which is what you want if you will change it:

git clone https://github.com/ianderso/german-archives-mcp
cd german-archives-mcp
uv sync
uv run german-archives-mcp   # stdio server, usually launched by the client

Either way the server speaks MCP over stdio, so you will normally let an MCP client start it rather than run it by hand.

Claude Desktop

{
  "mcpServers": {
    "german-archives": {
      "command": "uvx",
      "args": ["german-archives-mcp"]
    }
  }
}

A desktop app does not always inherit your shell's PATH. If the server fails to start because uvx cannot be found, give the full path that which uvx prints as the command.

Claude Code

claude mcp add german-archives -- uvx german-archives-mcp

Configuration

Nothing is required. A .env file in the directory the server starts in supplies anything the environment does not; only that directory is read.

An unusable value is reported on the first tool call as a not_configured result naming the variable.

How to read what comes back

• A DDB entry is a finding aid. The archive wrote it to say where a record is and roughly what it holds. It supports a research task (order the file, open the images), not a fact. Cite the archive and its signature (citation in ddb_item), and the archive's own page where the entry links one; keep the DDB id only as a finder.
• A DDB search covers descriptions, never page text, and only what archives have delivered. Many German archives deliver nothing, and most that do describe registers by volume, not by name. No hits is not a negative.
• archive matches the start of the holder's name, exactly as the DDB writes it, capitals and umlauts included: Stadtarchiv Düsseldorf, not stadtarchiv duesseldorf.
• Dates match by overlap. year_to: 1812 keeps a register for 1811-1816.
• A series heading has no record of its own. ddb_item then answers with its place in the hierarchy and its children (grouping_node).
• The DDB lists only the first ten images of a Landesarchiv item. labw_image reads every page; ddb_search and ddb_item give the permalink it takes (labw_permalink).
• A Baden Standesbuch is a civil copy of the parish register, 1810 to 1870; a Württemberg duplicate, 1808 to 1875. Before, look for the parish register itself; after, the civil registry office. A Baden book often covers several years and all three registers, and sometimes two confessions (confessions); a Württemberg volume holds one register (registers).
• labw_standesbuch finds books by the place the archive files them under. In Baden and J 386 the place is in the book's title; in F 901 and Wü 110 T 1 it is the parish heading above the volume (place). A village incorporated into a town is filed under both names (Niederrimsingen, Breisach am Rhein FR; Einsingen, Ulm UL), so a search for the town lists its villages too; place_named_first says which books are the place's own. Two places can share a name: Ulm in Württemberg and two Ulms in Baden. Spell the place as the archive does, umlauts included.
• has_images: false means the volume is described but not online. In Wü 110 T 1, {1754: nicht vorhanden} marks a duplicate that was never delivered.
• J 386 is film, not the original. The Reichssippenamt filmed the Jewish registers in 1943-1945; the originals are lost, so the film is the surviving witness. Cite the film's signature.
• An NRW search covers descriptions. Registers are described by Standesamt (or, before 1815, Mairie or Bürgermeisterei), register type and year: search Geburtsregister Barmen, not a person's name. Years match by overlap. Only the Landesarchiv's own units are read by nrw_image; other archives' digitised units come back with a DFG-Viewer link to open in a browser.
• page is the archive's image number ("Bild"), counted from 1, not the number written on the page. In NRW it is the image's place in the METS file, which the archive's viewer shows in the same order. A book's image numbers and its permalinks can fall out of step where an image was added later: in L 10 Nr. 510, Bild 119 is permalink …-380, and Bild 378 is …-377. labw_image reads the archive's own list, so either works.
• APERTUS files a register under its Bürgermeisterei or Standesamt. Search the office's name (Cochem finds Cochem-Land and Cochem-Stadt); rlp_search covers the units with images, mostly from the former government districts of Koblenz and Trier. Some units' images are only the register's name index; rlp_image passes on the archive's note.
• rlp_image's page is the page of the volume's PDF, one scan per page, often a two-page spread. The first page of a volume costs the archive a whole volume (6 to 27 MB in the registers seen), so read several pages of one volume rather than one page each of many; the volume stays on disk for GERMAN_ARCHIVES_PDF_CACHE_DAYS. The page saved is the archive's own JPEG, taken out of the PDF unchanged.
• LAGIS matches every word against the register's title and the places of its Standesamt district, so a village finds its Standesamt's registers. A year matches only a register whose title names it: a volume binding several years needs no year. LAGIS orders hits by relevance, 25 at most a page.
• hessen_image's page is the image's place in Arcinsys's viewer. Pages still under a protection period (an adoption, an acknowledged paternity) are left out of the online register, so an image's file name can be one ahead of its place; the citation gives both.
• Read the page. These registers are handwritten, mostly in German script. The image is the evidence; nothing in this server transcribes it.

Terms of use, and being a good guest

Deutsche Digitale Bibliothek. The DDB's developer documentation says its data is "publically accessible without any restrictions" (DDB-Backend / API) and that "using version 2 all endpoints are usable without an API key" (Differences between API versions 1 and 2). Its OpenAPI description says metadata delivered through the API is licensed CC0, while images carry each institution's own licence. Two older pages still describe a key: the DDBpro page on its interfaces and the general text of the OpenAPI description; the version 2 routes this server uses declare no security and answered without one on 2026-10-11. The former API terms page (/content/terms/api) now answers 404, and the DDB's portal sits behind an Anubis proof-of-work check, which this server never touches: it reaches only the API host. No rate limit is published. The API host has no robots.txt (it answers 404). If the DDB starts refusing keyless requests, the tools say key_required.

Landesarchiv Baden-Württemberg. Its terms (Nutzungsbedingungen auf einen Blick) put the online finding aids' metadata under CC0, mark each digitised item with its rights (the viewer shows the Public Domain Mark 1.0 for every fonds listed above, checked 2026-10-11), and ask every user to "Archivsignatur oder den Permalink zitieren". They also ask for a deposit copy of a book made with substantial use of the archive's material (§ 8 Abs. 9 Landesarchivgesetz). They say nothing about automated access. As a fact: robots.txt on www2.landesarchiv-bw.de allows Googlebot and Yahoo and disallows everyone else. The DDB's copy of the Landesarchiv's entries lists the images as CC BY 3.0 DE; the archive's own viewer marks them Public Domain, and labw_image reports the archive's mark.

Landesarchiv Nordrhein-Westfalen. Its terms for content and digitised records (Nutzungsbedingungen, PDF) put its finding aids under CC0 and its digitised records under CC BY-SA; digitised records "dürfen von der Webseite in der bereitgestellten Auflösung kostenlos heruntergeladen werden", and every reuse must cite the signature, and online the permalink. Each METS file repeats the licence. The terms say nothing about automated access. As facts: robots.txt on www.archive.nrw.de disallows the site's own /search/ page and administrative paths, not the /sufservice/ search service this server calls; www.landesarchiv-nrw.de has no robots.txt (404). No bot check was met on either host.

Landesarchivverwaltung Rheinland-Pfalz (APERTUS). APERTUS's help says that every digital version of archival material in it is free to download ("stehen in APERTUS zum kostenlosen Download zur Verfügung", Hilfe > Download), and the archive's fee page says downloads from APERTUS are free of charge (Dienstleistungen und Gebühren). Its decree on publication (Allgemeinverfügung zu Veröffentlichungsgenehmigungen, PDF) says records free for unrestricted use need no permission to publish, asks that the sources be cited (§ 10 Abs. 1 LArchBVO), politely asks for a copy, waives publication fees, and forbids unlawful, defamatory or extremist use, as APERTUS's disclaimer does. Nothing on automated access. As a fact: apertus.rlp.de has no robots.txt (404). No bot check was met; the search and unit pages answered without a cookie.

Hessisches Landesarchiv (Arcinsys Hessen, LAGIS). The archive says its digitised civil registers can be seen in Arcinsys free of charge and without registration, and that single pages may be downloaded as JPG (Recherche in digitalisierten Personenstandsnebenregistern); use of reproductions published online is free of charge (Kosten und Gebühren). No licence for the images is published: each METS file's licence field is empty, and Arcinsys's imprint notes only that some content is protected by copyright. LAGIS gives its catalogue data as CC0 and a citation form for each entry. Nothing on automated access was found in either. As facts: arcinsys.hessen.de/robots.txt names only a sitemap; digitalisate-he.arcinsys.de/robots.txt disallows everything for every agent; www.lagis-hessen.de/robots.txt disallows named crawlers and some search pages, not /de/api/. No bot check was met; no session, form state or script was needed.

What the client does. It sends one request at a time to each host, at least two seconds apart, and holds the searches that run on an archive's own database (the Archive in NRW portal's, APERTUS's and LAGIS's) at least ten seconds apart, and the opening of two APERTUS volumes, each of which makes the archive build and send a whole PDF, at least ten seconds apart. Two identical calls in flight share one request. Searches are cached for a day, DDB records for a week, and the Landesarchiv Baden-Württemberg's permalink answers and image lists, the Landesarchiv NRW's and Arcinsys's METS files and the signature look-ups for 30 days, so reading a second page of a book costs one request. An APERTUS volume is downloaded once and kept for three days (within a 1000 MB cap; both configurable), so its later pages cost none. A 429, a 5xx or a dropped connection gets one retry, honouring Retry-After. The User-Agent names the package, its version and this repository; no other client is imitated. Images are fetched at the size the archive's viewer shows, and written to your file, never cached.

Deliberately not here

• The DDB's portal, newspapers and user features. The portal is behind a bot check; favourites and saved searches need an account.
• Images on the holders' own sites. ddb_item and nrw_search return their links; this server fetches nothing outside its four hosts.
• Other states' registers, for now. Berlin's are on Ancestry, Saarland's and Bavaria's are not online at their state archives, and Thuringia's and Saxony's church books are on Archion. See docs/DESIGN.md.
• Resizing. rlp_image and hessen_image save the archive's JPEG as it is; making it smaller would need an imaging library.
• Matricula. Its church books span several countries and deserve a server of their own.
• Transcription. The page is read by a person, or by a tool built for German script.
• Working around bot checks. A site that answers with one is reported as blocked and left alone.

Security

Tool arguments are written by a model, and the model reads text this server does not control: finding-aid entries, image links, web pages. The server assumes that text can steer the model, and limits what a steered model can make it do.

• Which hosts. Eight, fixed: api.deutsche-digitale-bibliothek.de, www2.landesarchiv-bw.de, www.archive.nrw.de, www.landesarchiv-nrw.de, apertus.rlp.de, www.lagis-hessen.de, arcinsys.hessen.de and digitalisate-he.arcinsys.de. Any other host is refused before it is looked up, including addresses that come back in a response, such as an image link in a DDB record, a viewer link in an NRW hit, a PDF path in an APERTUS answer, an image in a METS file, or a redirect. A permalink redirect is read, never followed.
• Which addresses. Each connection is checked where it is made: a name that leads to a private, loopback, link-local, CGNAT, multicast, reserved or unspecified address, IPv4 or IPv6, is refused, and the connection goes to the address that was checked. Proxy settings in the environment are not used.
• How much. A JSON, HTML or XML answer over 10 MB, an image over 60 MB, or a volume PDF over GERMAN_ARCHIVES_PDF_CACHE_MB (and never over 500 MB), is refused as it streams in. A search returns at most 50 hits a page.
• What is sent. Search words have Solr's field, range and parameter syntax escaped; a place must be letters, spaces and a few punctuation marks; ids and signatures are matched against narrow patterns; an image is asked for only by a file name the archive's own list gave, and only if it has the archive's file-name shape; an NRW or Hessian image only by an address its METS file gave, on the archive's image server, ending in .jpg, with no .. and no query; an APERTUS volume only by the /Bild/ path the unit's own description gave, of the archive's file-name shape. A cached volume's file name is the unit's GUID, checked against its pattern.
• Which files. The four image tools each create one new file and never overwrite one; rlp_image also writes volume PDFs inside the cache folder, nowhere else. The bytes must be an image, judged by their first bytes, and a JPEG, since the file must end in .jpg or .jpeg. Never a hidden file or folder, never under ~/Library, and with GERMAN_ARCHIVES_DOWNLOAD_DIR set, never outside it, all judged after links are resolved. A refused download leaves nothing on disk.
• Site text is untrusted. Titles and descriptions reach the model verbatim. The server's instructions tell the model to treat that text as material to weigh, never as instructions; the model still decides, so review what it proposes to do.

To report a vulnerability, see SECURITY.md.

Development

uv sync --extra dev
uv run pytest                      # mocked with respx; never touches a site
uv run ruff check .
uv run ruff format --check .
uv run python -m tests.live_check  # paced calls to the live sites

The live check asks the sites what the recorded fixtures cannot: whether their answers still have the shape the server reads. See CONTRIBUTING.md for how the suite is organised, docs/API-NOTES.md for what was observed of each site and when, and docs/DESIGN.md for why the server is shaped this way.

Credits

The finding aids belong to the archives that wrote them, and the images to the archives that hold the records. The Deutsche Digitale Bibliothek is run by a network of German cultural institutions; the Landesarchiv Baden-Württemberg, the Landesarchiv Nordrhein-Westfalen, the Landesarchivverwaltung Rheinland-Pfalz and the Hessisches Landesarchiv are the state archives of Baden-Württemberg, North Rhine-Westphalia, Rhineland-Palatinate and Hesse; the Archive in NRW portal is run by the Landesarchiv NRW for the archives of the state; LAGIS is run by the Hessisches Landesamt für geschichtliche Landeskunde; Arcinsys was built by the Hessisches and the Niedersächsisches Landesarchiv.

License

MIT.