The DHLAB MCP Server provides tools for searching and analyzing the National Library of Norway's digital collections, enabling text analysis, linguistic research, and corpus statistics.
Text Search: Search the digital collection for books, newspapers, or journals by keyword, with optional filters for year range and media type.
NGram Frequency Analysis: Analyze and compare word frequency trends over time across book or newspaper corpora (1810–2020).
Concordance (HTML): Retrieve HTML-formatted in-context occurrences of a word within a specific document, useful for display purposes.
Concordance (Structured): Retrieve clean, structured concordance data (before/target/after fields) for programmatic analysis of word contexts.
Collocations: Discover words that frequently appear near a target word within a document, with a configurable context window.
Word Form Lookup: Retrieve all morphological forms of a given Norwegian word.
Word Lemma Lookup: Find the base/dictionary form (lemma) of a given Norwegian word.
Image Search: Search for images in the digital collection by keyword and optional year range.
Corpus Statistics: Retrieve statistical metadata about a set of documents identified by their URNs.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@DHLAB MCP Server search for texts about Ibsen from 1900-1950"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server providing access to DHLAB (National Library of Norway Digital Humanities Lab) functionality through the Model Context Protocol.
This server exposes tools for:
Text search: Search the National Library's digital text collection
NGram analysis: Analyze word frequency trends over time
Concordance: Find word contexts in documents
Collocations: Discover words that appear together
Word lookup: Look up Norwegian word forms and lemmas
Image search: Search for images in the digital collection
Corpus statistics: Get information about document collections
Related MCP server: gallica-mcp
This project uses uv, which can be installed with:
# On macOS and Linux.
curl -LsSf https://astral.sh/uv/install.sh | shClone and install:
git clone https://github.com/marksverdhei/dhlab-mcp.git
cd dhlab-mcp
uv sync --devOr install directly:
pip install git+https://github.com/marksverdhei/dhlab-mcp.gitAdd the MCP server to your Claude Code configuration:
# inside the repo directory:
claude mcp add --transport stdio dhlab -- uv --directory $PWD run dhlab-mcpor under user scope:
claude mcp add --scope user --transport stdio dhlab -- uv --directory $PWD run dhlab-mcpVerify the server is added:
claude mcp listThe DHLAB tools will then be available in your Claude Code sessions.
You can also run the server directly for testing:
dhlab-mcpOr in development mode:
uv run dhlab-mcpTo run the MCP server as a local HTTP API on a custom port:
# Run on default port 8000
dhlab-mcp --transport http
# Run on a custom port
dhlab-mcp --transport http --port 9000
# Run on a specific host and port
dhlab-mcp --transport http --host 0.0.0.0 --port 8080The server supports the following transport options:
stdio (default): Standard input/output for CLI integration
http: Streamable HTTP transport (recommended for network access)
sse: Server-Sent Events transport (legacy, for backward compatibility)
Once running, the HTTP server will be available at http://<host>:<port>/mcp/.
search_textsSearch for texts in the digital collection.
{
"query": "ibsen",
"limit": 10,
"from_year": 1900,
"to_year": 1950,
"media_type": "aviser" # or "bøker", "tidsskrift"
}ngram_frequenciesGet word frequency trends over time.
{
"words": ["frihet", "demokrati"],
"corpus": "bok", # or "avis"
"from_year": 1810,
"to_year": 2020
}find_concordancesFind word contexts in a document (returns HTML-formatted text).
{
"urn": "URN:NBN:no-nb_digibok_2008051404065",
"word": "Norge",
"window": 25
}Output format: HTML-formatted concordance with <b> tags highlighting matches.
word_concordanceFind word contexts with structured output (no HTML formatting).
{
"urn": "URN:NBN:no-nb_digibok_2008051404065",
"word": "Norge",
"window": 12
}Output format: Clean structured data with separate fields:
dhlabid: Document identifier
before: Text before the matched word
target: The matched word itself
after: Text after the matched word
Use cases:
Use find_concordances for display/UI (HTML-formatted)
Use word_concordance for analysis/processing (structured data)
find_collocationsFind words that appear near the target word.
{
"urn": "URN:NBN:no-nb_digibok_2008051404065",
"word": "frihet",
"window": 5
}lookup_word_formsLook up different forms of a Norwegian word.
{
"word": "løpe"
}lookup_word_lemmaLook up the lemma (base form) of a word.
{
"word": "løper"
}search_imagesSearch for images in the collection.
{
"query": "Oslo",
"limit": 10,
"from_year": 1900,
"to_year": 1950
}get_corpus_statisticsGet statistics about a set of documents.
{
"urns": ["URN:NBN:no-nb_digibok_2008051404065"]
}For development, install with:
uv sync --devRun tests:
pytestFormat code:
ruff format src/ tests/DHLAB is a Python library for qualitative and quantitative analyses of digital texts from the National Library of Norway's collection. For more information, visit:
See LICENSE file.
Find collocations (words that appear near the target word) in a document.
Args: urn: URN identifier for the document word: Target word to find collocations for window: Size of context window (default: 5) limit: Maximum number of collocations to return (default: 100)
Returns: JSON string containing collocation statistics
Inputs: urn (required), word (required), window, limit. Returns: result
Find concordances (contexts) for a word in a specific document.
Args: urn: URN identifier for the document word: Word to find concordances for window: Number of words before and after the match (default: 25) limit: Maximum number of concordances to return (default: 100)
Returns: JSON string containing concordance results
Inputs: urn (required), word (required), window, limit. Returns: result
Get statistical information about a corpus of documents.
Args: urns: List of URN identifiers for documents
Returns: JSON string containing corpus statistics
Inputs: urns (required). Returns: result
Look up different forms of a Norwegian word.
Args: word: The word to look up
Returns: JSON string containing different word forms
Inputs: word (required). Returns: result
Look up the lemma (base form) of a Norwegian word.
Args: word: The word to look up
Returns: JSON string containing lemma information
Inputs: word (required). Returns: result
Get word frequency trends over time using NGram analysis.
Args: words: List of words to analyze corpus: Corpus type. Options: 'bok' (books), 'avis' (newspapers). Default: 'bok' from_year: Start year (default: 1810) to_year: End year (default: 2020) smooth: Smoothing parameter for the frequency curve (default: 1)
Returns: JSON string containing frequency data over time
Inputs: words (required), corpus, from_year, to_year, +1 more. Returns: result
Search for images in the National Library's digital collection.
Args: query: Search query string limit: Maximum number of results (default: 10) from_year: Start year (optional) to_year: End year (optional)
Returns: JSON string containing image search results with URLs
Inputs: query (required), limit, from_year, to_year. Returns: result
Search for texts in the National Library's digital collection.
Args: query: Search query string limit: Maximum number of results to return (default: 10) from_year: Start year for search period (optional) to_year: End year for search period (optional) media_type: Type of media to search. Options: 'digavis' (newspapers), 'digibok' (books), 'digitidsskrift' (journals). Default: 'digavis'
Returns: JSON string containing search results with metadata
Inputs: query (required), limit, from_year, to_year, +1 more. Returns: result
Find concordances with structured output (no HTML formatting).
Returns clean format with separate before/target/after fields instead of HTML-formatted text. This is useful for programmatic analysis where you need the matched word separated from context.
Args: urn: URN identifier for the document word: Word to find concordances for window: Number of words before and after the match (default: 12, max: 24) limit: Maximum number of concordances to return (default: 100)
Returns: JSON string containing structured concordance results with fields: - dhlabid: Document identifier - before: Text before the matched word - target: The matched word itself - after: Text after the matched word
Inputs: urn (required), word (required), window, limit. Returns: result
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 9 tools
Most tools have distinct purposes, but find_concordances and word_concordance have significant overlap as both find concordances for words in documents, differing mainly in output format. This could cause confusion for an agent deciding which to use. Other tools like find_collocations, ngram_frequencies, and search functions are clearly differentiated.
All tool names follow a consistent snake_case pattern with clear verb_noun structure (e.g., find_collocations, lookup_word_forms, search_images). The naming is predictable and readable throughout the set, with no mixing of conventions or styles.
With 9 tools, this server is well-scoped for digital humanities text analysis. The count is appropriate, covering key operations like searching, concordancing, collocation analysis, and statistical queries without being overwhelming or too sparse for the domain.
The toolset covers core digital humanities workflows including text search, image search, concordancing, collocation analysis, and corpus statistics. Minor gaps exist, such as no direct document retrieval or metadata lookup tools, but agents can likely work around these using the provided search and analysis functions.
Multi-engine scholarly research server for search, traversal, full text, and reading lists.
Read-only search and retrieval of published books, articles and authors, with public catalog links.
Search biomedical papers, inspect publication records, and traverse citation or semantic graphs.
Search the Wayback Machine and IA library (40M+ items), fetch snapshots, item metadata, and text.