search-letters

You can search a historical text corpus without a backend API by fetching and filtering precomputed static JSON endpoints client-side. This approach retrieves letter summaries, snippets, and semantic relationships directly in the browser for fast exploration.

You filter historical Danish letters by date, sender, and recipient using the letter-summaries.json endpoint. This static file provides the necessary metadata for client-side filtering across the 665-letter WW1 correspondence collection.

Yes, you can perform semantic similarity searches on a static document collection by utilizing a precomputed related-letters.json file. This file contains cosine-similarity based recommendations that allow client-side thematic exploration without server-side processing.

The best way to preview full-text letters before fetching the entire document is by using the search-snippets.json endpoint. It provides 200-character snippets for fast triage, allowing you to quickly scan matches before retrieving full markdown pages.

Yes, this semantic search approach works for browsing thematic clusters in a historical corpus by loading the precomputed topic-clusters.json file. It enables thematic browsing and retrieval of representative samples entirely client-side.

No, you do not need a server to retrieve full markdown pages from a static JSON corpus. The Skill supports content negotiation to fetch individual markdown letter pages directly from static endpoints alongside the bulk plain-text search-corpus.json file.