semanticscholar-skill

Semantic Scholar Search Workflow

Search academic papers via the Semantic Scholar API using a structured 4-phase workflow.

Critical rule: NEVER make multiple sequential Bash calls for API requests. Always write ONE Python script that runs all searches, then execute it once. All rate limiting is handled inside s2.py automatically.

Phase 1: Understand & Plan

Parse the user's intent and choose a search strategy:

Decision Tree

Default to search_bulk(). Per Semantic Scholar's own docs, bulk search is preferred over relevance search for most cases because relevance search is more resource-intensive. Use search_relevance() only when you need TLDR fields or author/citation details inline.

Query Construction Rules

• Ambiguous terms (e.g., "stem cells" could mean mesenchymal or stem-like T cells): Use build_bool_query() with exact phrases and exclusions
  • Example: build_bool_query(phrases=["stem-like T cells"], required=["CD4", "TCF7"], excluded=["mesenchymal", "hematopoietic stem cell"])
• Multi-context queries (e.g., "topic X in cancer AND autoimmunity"): Plan separate searches, deduplicate with deduplicate()
• Broad topics: Use search_relevance() with filters (year, venue, fieldsOfStudy, minCitationCount)

Plan Filters

Checkpoint: Before proceeding, verify: (1) search strategy matches user intent, (2) filters are appropriate, (3) query is specific enough to avoid irrelevant results.

Phase 2: Execute Search

Write ONE Python script that begins with the standard prelude below, then runs all searches:

# --- Standard prelude (use in every script) ---
import sys, os, glob
_candidates = [

Save to /tmp/s2_search.py, then run with python3 /tmp/s2_search.py in a single Bash call. Rate limiting, retries, and backoff are automatic inside s2.py.

No API key: The skill works without S2_API_KEY. When the key is absent or invalid, s2.py automatically switches to unauthenticated mode (no x-api-key header) and widens the request gap to 5 s. Per S2 docs, anonymous calls share a global 1000 req/s pool across all unauthenticated users and can be "further throttled during periods of heavy use" — so a conservative 5 s gap protects against the heavy-use throttling, even though the steady-state pool is generous. If you still see sustained 429s, raise _MIN_GAP to 10 s. Keep max_results ≤ 30 per search and combine fewer searches per script. S2 recommends including an API key on every request — get one at https://www.semanticscholar.org/product/api#api-key-form.

Checkpoint: Verify the script ran successfully (no exceptions) and returned results. If 0 results, broaden the query or relax filters before presenting.

Worked Examples

Each example below assumes the standard prelude from Phase 2 is at the top of the script.

Example 1: Author workflow — "Find papers by Yann LeCun on self-supervised learning"

Example 2: Citation chain with intent — "Who cited the Transformer paper and how did they use it?"

Example 3: Multi-seed recommendations with BibTeX export — "Find papers like these two but not about NLP"

Phase 3: Summarize & Present

• Use format_results() for consistent output (summary table + top-10 details)
• If user's language is Chinese, present summaries in Chinese
• Always note total results count and search strategy used
• Highlight most relevant papers based on the user's specific question

Phase 4: User Interaction Loop

After presenting results, always offer these options:

1. Translate — titles/summaries to Chinese (or other language)
2. Details — full abstract for specific paper numbers
3. Refine — narrow or expand search with different terms/filters
4. Similar — find papers similar to a specific result (find_similar())
5. Citations — who cited a specific paper and how (get_citations() + format_citations() for intent labels)
6. Export — save results via export_bibtex(), export_markdown(), or export_json()
7. Done — end search session

Loop until user says done. Each follow-up uses the same single-script pattern.

Additional Resources

• S2folks GitHub — Official Semantic Scholar code examples: https://github.com/allenai/s2-folks
• Postman Collection — No-code API testing: linked from https://www.semanticscholar.org/product/api/tutorial
• API Documentation — Full endpoint reference: https://api.semanticscholar.org/

API Quick Reference

Helper Module (s2.py)

Use the standard prelude from Phase 2 at the top of every script. Then call any of the functions below — the module's docstring (help(s2) or read s2.py) lists each by phase with one-line summaries.

Paper Search Functions

Author Functions

Filter Parameters (kwargs)

snake_case kwargs are translated to S2 camelCase params automatically (fields_of_study → fieldsOfStudy, min_citations → minCitationCount, publication_date → publicationDateOrYear, pub_types → publicationTypes, open_access → openAccessPdf). Use snake_case here.

year, publication_date, venue, fields_of_study, min_citations, pub_types, open_access

• year: "2020-", "-2019", "2016-2020"
• publication_date: "2024-01-01:2024-06-30" (YYYY-MM-DD range, open-ended OK)
• pub_types: Review, JournalArticle, Conference, ClinicalTrial, MetaAnalysis, Dataset, Book, CaseReport, Editorial, LettersAndComments, News, Study, BookSection

Boolean Query Syntax (bulk search only)

Use build_bool_query(phrases, required, excluded, or_terms, fuzzy, proximity) to construct safely.

• fuzzy: list of (term, edit_distance) tuples
• proximity: list of (phrase, word_distance) tuples

Output Functions

Supported ID Formats

DOI:10.1038/..., ARXIV:2106.15928, PMID:19872477, PMCID:PMC2323569, CorpusId:215416146, ACL:2020.acl-main.447, DBLP:conf/acl/..., MAG:3015453090, URL:https://...

Paper Fields

Default: title,year,citationCount,authors,venue,externalIds,tldr

Additional: corpusId (integer, S2 secondary ID), url (S2 paper page link), abstract, references, citations, openAccessPdf, publicationDate, publicationVenue, fieldsOfStudy, s2FieldsOfStudy, journal, isOpenAccess, referenceCount, influentialCitationCount (influential citations only), citationStyles, embedding, textAvailability

externalIds object contains: ArXiv, MAG, ACL, PubMed, Medline, PubMedCentral, DBLP, DOI

Author fields: name, affiliations, paperCount, citationCount, hIndex, homepage, externalIds, papers

Minimize fields. Per the official S2 tutorial: "Avoid including more fields than you need, because that can slow down the response rate." Only add abstract, references, or citations when the user explicitly needs them.

sort Parameter Values (bulk search only)

The sort kwarg accepts only these three values:

Recommendations Limits

find_similar() and recommend() return at most 500 papers per call (limit max = 500).

Datasets API Functions

For bulk download of full S2 datasets (papers, authors, abstracts, embeddings, etc.):

Available dataset names (pass as dataset_name):

All datasets are delivered as JSON Lines (one record per line). The diffs response contains update_files (insert/replace by primary key) and delete_files (remove from dataset).

Rate Limiting

s2.py adapts automatically based on whether S2_API_KEY is set:

S2 recommends including an API key on every request, even for endpoints that work anonymously — it gives you a dedicated quota, a smoother experience under load, and better support if you need help. The introductory 1 req/s key can be raised on request. Get one at https://www.semanticscholar.org/product/api#api-key-form

The anonymous 1000 req/s pool is generous in steady state, but the docs explicitly warn it can be throttled hard during heavy use — that is why _MIN_GAP defaults to 5 s without a key, not the 1 ms a 1000 req/s budget would technically allow. If your workload still hits sustained 429s, set _MIN_GAP = 10.0 in s2.py or get a key. The 1 req/s key budget is cumulative across all endpoints, so chained calls (e.g. get_paper → get_citations) count separately.

Bulk Search Response Structure

search_bulk() returns a list of papers already unpaginated. Internally the raw response has:

s2.py handles token pagination automatically — you only see the final flat list.

Troubleshooting