SNEWPapers Research
Server Details
Search extracted newspaper articles from the Library of Congress, 1736-1963. Subscription required.
- Status
- Healthy
- Uptime
- 5.6% over the last 24 hours
- OAuth
- Works in Glama
- Last Tested
- Transport
- Streamable HTTP · MCP 2025-11-25
- URL
- ChatGPT
- –
- Claude
- SNEWPapers
TDQS
Scored across 11 tools
Most tools have clearly distinct purposes: search vs. facets vs. component retrieval vs. corpus stats vs. title metadata vs. billing/docs. The only mild overlap is between research_labels (label definitions) and research_vocabulary (types/themes/keywords), which both provide metadata but at different levels of detail. Singular/plural pairs (component/components, title/titles) are explicitly differentiated by their descriptions.
Every tool follows the same predictable research_<noun> convention with consistent snake_case. Singular/plural variants are used deliberately and consistently (component vs. components, title vs. titles). No mixed styles or vague verbs.
Eleven tools is well within the typical 3-15 range and each tool earns its place: search, exploration, retrieval, metadata, billing, and documentation. No tool appears redundant or padding the surface.
The surface covers the full research lifecycle: corpus overview, faceted exploration, search, single and batch component retrieval, title discovery, label/vocabulary reference, and token accounting. Minor gaps exist—no explicit citation/export or saved-collection tool—but these are handled by the guide and workarounds, so agents are unlikely to hit dead ends.
Publisher details
- Operator
- SNEWPapers, operated by Brett Shinnebarger · Publisher source
- Operator website
- https://snewpapers.com
- Vendor relationship
- First-party · Publisher source
- Documentation
- https://snewpapers.com/connect
- Privacy policy
- https://snewpapers.com/privacy
- Trust center
- Not available
- Restrictions
- A SNEWPapers subscription or promo grant is required; usage is metered in research tokens. Sign in with Google or email through OAuth. No admin approval, regional limit or custom OAuth app is needed. · Publisher source
Available Tools
11 toolsresearch_account_tokensYour token balanceARead-onlyIdempotentInspect
Free. Your balance, the grant that resets each period, purchased tokens that carry over, this period's spend by operation, and your recent charges. Every metered result already carries tokens_charged and tokens_remaining; call this when the user asks how many tokens are left or a call was refused for insufficient tokens.
Inputs: no named parameters. Returns: grant, recent, balance, period_end, +5 more
research_componentOne component, wholeARead-onlyIdempotentInspect
The component a hit points at (uuid or event id): every event with its summary and structured fields, the evidence behind the quality tier, links to the Library of Congress page PDFs and to snewpapers.com (1 token). include_text=true adds the full text: 3 tokens, charged once per component - be selective, never page through full texts. component_id is a uuid or {uuid}_evt_N; the response echoes the requested event index so a client can highlight it. Caution components return only a description and links while guarded=true; for research the better experience is guarded=false - ask the researcher once, at the start of the session, then pass it on every call. If your own content rules cannot handle such historical text, do not fetch unguarded - give the researcher the site link. include_boxes=true (free) adds the component's rectangles per page as fractions of the page image, for drawing it.
Inputs: component_id (required), guarded, include_text, include_boxes. Returns: city, lccn, uuid, year, +27 more
research_componentsUp to 25 components at onceARead-onlyIdempotentInspect
A search page's worth of components in one call, charged in one transaction, at the single-call prices: without text 1 per component; with text 3 per component the first time this account reads it and 0 after. Priced before anything is fetched and refused whole if the balance cannot cover it, so check tokens_remaining first when asking for many full texts. Unknown ids come back in missing and cost nothing.
Inputs: query (required). Returns: missing, components, tokens_charged, tokens_remaining
research_corpusWhat the corpus isARead-onlyIdempotentInspect
Size, span (1736-1963), decade histogram, quality and caution shares, and coverage of Chronicling America (English-language titles are the target). Call once before searching so a negative result means something. Free.
Inputs: no named parameters. Returns: docs, issues, states, titles, +12 more
research_facetsAggregations over the matching componentsARead-onlyIdempotentInspect
The cheap exploration call (2 tokens): what labels, papers, states, years and terms the matching set holds, as flat or nested (depth 2) facets; returns no hits. Example: {"query": {"filters": {"start_date": "1832-06-01", "end_date": "1832-12-31"}, "text": {"must_have_keywords": "cholera"}, "flat_facets": ["states", "month_histogram"]}}. Everything under one query key; text matching under text as strings; term_filters / any_of are lists of {"field": ..., "values": [...]} objects; an unknown key is an uncharged 422 with a hint. Facets count matches, not a ranking: anchor with must_have_phrase or must_have_keywords and filters; should_* and semantic_text only widen. At least one of flat_facets / nested_facets is required. The presets (label counts, place x era coverage, hotspots, a field census) are in research_guide section "api".
Inputs: query (required). Returns: mode, corpus, facets, total_docs, +3 more
research_guideThe research method (the skill), in sectionsARead-onlyIdempotentInspect
How to research this corpus well - the procedure, the text-field rules, what each call costs, caution handling, citing. Free. Call it with section "start" FIRST, once per conversation, before any search: section "start" (the default) is the part to read first and lists the others; "api" has every field and filter, "caveats" the traps, "examples" twelve worked investigations, "display" how to present results, "research-folder" the per-topic folder. Returns markdown, never search results.
Inputs: section. Returns: The research method itself: the same skill a Claude Code or Gemini CLI user ins…
research_labelsThe labels and their fieldsARead-onlyIdempotentInspect
The 32 served labels (genres), each with its embedding rule (never / conditional), whether it carries themes, its field prefix and structured field list. Static; read once. A never-embed label cannot be found by semantic_text. Free.
Inputs: no named parameters. Returns: labels
research_searchOne page of 25 component summariesARead-onlyIdempotentInspect
Keyword, phrase, semantic or hybrid search with filters: one page of 25 component summaries with their lists and structured fields, never full text (1 token per page). Example: {"query": {"filters": {"start_date": "1918-09-01", "end_date": "1919-06-30"}, "text": {"must_have_keywords": "influenza", "semantic_text": "the epidemic closes schools"}, "page": 1}}. hit_shape=compact halves a page (identity + summary only). Page 1 returns a cursor; send it back with page=2..; a search stops at page 10 (narrow with research_facets); only issue browse (filters.issue_uuid, sort pages:asc, no text) pages a whole issue. Keep a hit's event_id ({uuid}_evt_N) for research_component. Caution hits carry only a description and a link while guarded=true; for research the better experience is guarded=false on every call, after asking the researcher once at the start. filters.created_after / updated_after return only what is new since your last pass.
Inputs: query (required). Returns: hits, mode, page, corpus, +7 more
research_titleOne title in depthARead-onlyIdempotentInspect
One paper (by lccn): processed issues per year against the Library of Congress span, so a missing year inside the span means "exists, not processed"; label and tier mix. With year, the processed issue dates in that year. Free.
Inputs: lccn (required), year. Returns: year, title, labels, quality_tiers, +2 more
research_titlesNewspaper titles with coverageBRead-onlyIdempotentInspect
Which papers exist, which are processed, and how much: filter by state (full name, e.g. "virginia"), city, county, ethnicity (the values the ethnicity facet returns, e.g. "African American", "German") or a name (q, contains, case-insensitive); returns total, refreshed_at and the page of titles (limit/offset); county, ethnicity, or a name (q; a masthead can span several lccns, each row carries its own). Untouched titles are listed with processed_issues=0. The lccn is the key used everywhere else. Free.
Inputs: q, city, limit, state, +5 more. Returns: total, titles, refreshed_at
research_vocabularyWhat you can filter onBRead-onlyIdempotentInspect
Types and themes (whole) and the top keywords for a label, optionally narrowed by lccn, state and years. Call before filtering on a term you have not seen. Free.
Inputs: lccn, size, label, state, +2 more. Returns: label, types, themes, keywords, +1 more
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Related MCP Connectors
Search LOC digital collections, Chronicling America newspapers (full OCR), and LC Subject Headings.
Chronicling America MCP — full-text search of ~150 years of digitized U.S. newspapers…
Search Virgin Islands newspapers page by page, 1865 to 2001, with the gaps in every run named.
Search magazine articles and public previews, with publisher attribution and subscription links.
Related MCP Servers
- AlicenseNot gradedqualityBmaintenanceEnables full-text search of digitized historical U.S. newspapers from the Library of Congress Chronicling America collection, returning publication metadata and page-scan image URLs with filters for date range and state.177 npmMIT
- AlicenseAqualityCmaintenanceEnables Claude to search, browse, and read more than 20 million digitized historical US newspaper pages from the Library of Congress's Chronicling America archive, covering the 1700s through 1963. Queries can be filtered by date range, state, newspaper, or front pages only, and results include full OCR text with citations and links back to each scan and PDF.4MIT
- AlicenseNot gradedqualityBmaintenanceSearch LOC digital collections, browse Chronicling America newspapers with full OCR text, and look up LC Subject Headings via MCP.466 npm6Apache 2.0
- AlicenseAqualityCmaintenanceMCP server for searching and retrieving full-text pages from the Library of Congress, including newspapers, books, and manuscripts, via the loc.gov API.3Apache 2.0
Glama MCP Gateway
Add one secure layer between your agents and this server.