Data Sources & Coverage
ra-mcp draws on two kinds of data sources: live Riksarkivet HTTP APIs (the core search/browse/viewer tools) and local LanceDB datasets that ship with the optional dataset modules.
Live HTTP APIs
The core modules connect to several Riksarkivet APIs:
| API | Endpoint | Purpose |
|---|---|---|
| Search API | data.riksarkivet.se/api/records |
Full-text search across transcribed documents |
| ALTO XML | sok.riksarkivet.se/dokument/alto |
Structured page transcriptions with text coordinates |
| IIIF | lbiiif.riksarkivet.se |
High-resolution document images and collection manifests |
| OAI-PMH | oai-pmh.riksarkivet.se/OAI |
Document metadata and collection structure |
| Bildvisaren | sok.riksarkivet.se/bildvisning |
Interactive image viewer (links provided in results) |
Most of this data comes from the Riksarkivet Data Platform, which hosts AI-transcribed materials from the Swedish National Archives. HTR re-transcription is delegated to the HTRflow Gradio Space.
Additional resources: Förvaltningshistorik (semantic search, experimental).
Local LanceDB datasets
The optional dataset modules query local LanceDB tables rather than a live API — they are full-text/vector indexes bundled with the dataset packages. They load only when lancedb is installed, and each adds its own namespaced search tools (see Module System). Scale figures below are reported by the modules themselves.
| Dataset (module) | Coverage | Tools |
|---|---|---|
SDHK + MPO (diplomatics) |
44,000+ medieval charters; 23,000+ parchment fragments | search_sdhk, search_mpo, view_sdhk, view_mpo |
SBL (sbl) |
9,400+ biographical articles (Svenskt biografiskt lexikon) | search_sbl, view_sbl_article, load_sbl_article |
Sjömanshus (sjomanshus) |
688,000+ seamen's records (voyages, registrations), 1700s–1900s | search_liggare, search_matrikel |
Filmcensur (filmcensur) |
60,000 film censorship records, 1911–2011 | search_filmreg |
Rosenberg (rosenberg) |
66,000 historical places (geographical lexicon) | search_rosenberg |
Court records (court) |
Domboksregister (Västra härad 1611–1730), Medelstad (1668–1750) | search_domboksregister, search_medelstad |
Aktiebolag (aktiebolag) |
Joint-stock companies 1901–1935 (~12.5K companies, ~49K board members) | search_bolag, search_styrelse |
Fältjägare (faltjagare) |
Jämtland field regiment soldiers 1645–1901 (~43K) | search_faltjagare |
Suffrage (suffrage) |
Women's suffrage records (Rösträtt petition 1913–1914, FKPR 1911–1920) | search_rostratt, search_fkpr |
Specialsök (specialsok) |
Flygvapen, fångrullor, kurhuset, press, video datasets | search_flygvapen, search_fangrullor, search_kurhuset, search_press, search_video |
DDS church records (dds) |
~2.5M births, deaths, marriages (1600s–1900s) | search_fodelse, search_doda, search_vigsel |
Wincars (wincars) |
Norrland vehicle registrations 1916–1972 (~1.5M across 5 counties) | search_wincars |
SJ railway (sj) |
Properties (198K JUDA) and technical drawings (118K FIRA/SIRA) | search_juda, search_ritningar |
TORA (tora) |
51K settlements with coordinates (historical-place geocoding) | search_tora |
In addition, the optional label module is not a dataset but an integration that imports pages to a Label Studio instance for human annotation (import_to_label_studio).
Archive Coverage
The archive has three access tiers — not all materials are searchable the same way:
| Tier | Tool | Coverage |
|---|---|---|
| Metadata catalog | search_metadata |
2M+ records — titles, names, places, dates |
| Digitised images | browse_document (links) |
~73M pages viewable via bildvisaren |
| AI-transcribed text | search_transcribed |
~1.6M pages — currently court records (hovrätt, trolldomskommissionen, poliskammare, magistrat) from 17th-18th centuries |
Church records, estate inventories, and military records are typically cataloged and often digitised, but NOT AI-transcribed.
Transcription Quality
The AI-transcribed text was produced by HTR (Handwritten Text Recognition) and OCR models. These transcriptions are not perfect — they contain recognition errors including misread characters, merged or split words, and garbled passages, especially in older or damaged documents.
This has a direct impact on search: an exact search for Stockholm will miss documents where the transcription reads Stockholn or Stookholm due to recognition errors. Always use fuzzy search (~) to compensate — stockholm~1 catches common misreads and significantly increases the number of hits.
The Plugin Model
ra-mcp is one piece of a larger ecosystem. Multiple MCP servers can be connected to the same AI client:
graph LR
client["AI Client\n(Claude)"]
client --> ramcp["ra-mcp\nSearch, browse, HTR, viewer, PDF,\nguides, dataset modules"]
client --> htrflow["htrflow-mcp\nStandalone HTR\n(alternative)"]
client --> other["other servers\nAny MCP-compatible tool"]Together with external tools, they enable a complete research workflow: search the archives, read transcriptions, re-transcribe pages that need better OCR/HTR, and view original documents — all from within a single AI conversation.