CitedEvidenceResearch APIs

CEDOA Files API

Fulltext, XML, stored PDFs, and downloadable PDF links.

Every CEDOA result has at least one eligible content source: archived fulltext, XML, a stored PDF, or a ranked downloadable PDF link. Metadata-only records are excluded. Not every paper has every format, and external links may later fail or expire.

Access and first request

Enable Files on your existing account API key. Use the same Bearer header as Metadata and Research MCP. The base URL is https://scholar.citedevidence.com/v1/research-corpus.

curl -sS -H "Authorization: Bearer $CE_API_KEY" \
  "https://scholar.citedevidence.com/v1/research-corpus/list?limit=5&has_fulltext=true"

Choose a returned work ID, inspect /manifest/{id}, then request an available format. In the Playground, the Files presets accept that work ID.

Content formats

Fulltext: has_fulltext
Archived text, streamed as text/plain. An abstract is not fulltext.
XML: has_xml
Archived XML, streamed as application/xml.
Stored PDF: has_archived_pdf
PDF bytes held in storage, streamed as application/pdf.
PDF link: has_pdf_link
An eligible HTTP(S) PDF source. This is not a stored PDF. Check the manifest for the provider, download_url, access status, and licence.

Read endpoints

  • GET/v1/research-corpus

    API index and route map.

  • GET/v1/research-corpus/list?limit=5

    Incremental feed of content-backed records.

  • GET/v1/research-corpus/search?q=climate&limit=5

    Search content-backed papers, excluding metadata-only records.

  • GET/v1/research-corpus/manifest/{id}

    Inspect eligible sources, provider rank, health, checksums, licences, and provenance.

  • GET/v1/research-corpus/text/{id}

    Stream archived fulltext. Returns 404 when no eligible text is available.

  • GET/v1/research-corpus/texts/{ids}

    Up to 100 comma-separated work IDs; unavailable texts are null in their result position.

  • GET/v1/research-corpus/xml/{id}

    Stream archived XML, decompressing storage gzip. Returns 404 when no eligible XML is available.

  • GET/v1/research-corpus/pdf/{id}

    Try stored PDFs first. If only an eligible external PDF link is available, return a 307 redirect. Returns 404 when neither is available.

  • GET/v1/research-corpus/stats

    Cached coverage measurements. Format counts overlap and must not be summed; missing measurements can be null.

Filters and pagination

  • limit: 1 to 1000; default 100.
  • q: search query for /search.
  • has_fulltext, has_xml, has_pdf: boolean filters. has_pdf includes stored PDFs and PDF links; it is not a stored-PDF-only filter.
  • provider: openalex, unpaywall, semantic_scholar, s2orc, pmc, europe_pmc, arxiv, crossref, pdf_derived, or other.
  • from_year / to_year: publication year bounds.
  • license: source licence pattern, for example cc-by% (URL-encode the percent sign as %25). Labels retain their source spelling.

For incremental pagination, pass next_indexed_after as indexed_after and next_indexed_after_id as indexed_after_id together. The ID prevents skipping papers that share an indexing timestamp. Keep the original filters.

File delivery and rights

Do not forward your API key to an external PDF host. Inspect the manifest and fetch external links separately without CitedEvidence credentials. The Playground stops redirects and offers stored binary responses as downloads.

A content source is not a redistribution or AI-training licence. Inspect source licences and access terms. A listed PDF link may not have been freshly download-verified.

Errors and quotas

401: missing or invalid key. 403: Files disabled. 404: work or requested format unavailable. 502: a stored artifact could not be read. 503: service unavailable. On 429, respect Retry-After. Files has a separate use-case quota, visible in Account API keys.