Get all citation papers for a specific paper using the Semantic Scholar Graph API. Use this skill when a user asks to find, list, or export papers that cite a specific paper (via arXiv ID, DOI, Semantic Scholar ID, etc.).
This skill enables Codex to retrieve the complete list of papers that cite a specific target paper (Citations) using the Semantic Scholar Graph API. It handles pagination automatically to ensure all citing papers are retrieved, even for highly cited works.
When a user asks for papers that cite a specific paper, follow these steps:
Extract the paper identifier from the user's request. Semantic Scholar supports several formats. You must format the ID correctly:
ARXIV:{id} (e.g., ARXIV:2509.16117)10.{id} (e.g., 10.1038/nrn3241){40-character hex string}MAG:{id}ACL:{id}PMID:{id}CorpusID:{id}Use the bash tool with curl to call the Semantic Scholar API.
Endpoint: GET https://api.semanticscholar.org/graph/v1/paper/{paper_id}/citations
Required Query Parameters:
fields: Comma-separated list of fields. Always request at least: title,authors,year,url. Other useful fields: venue,citationCount,abstract,isOpenAccess.limit: Set to 1000 (the maximum allowed) to minimize API calls.offset: Start at 0.Bash Command Example:
curl -s "https://api.semanticscholar.org/graph/v1/paper/ARXIV:2509.16117/citations?fields=title,authors,year,url&limit=1000&offset=0"
Analyze the JSON response from the API.
data array and an offset integer.data array is exactly equal to your limit (e.g., 1000), there are likely more citations to fetch.offset parameter by the limit (e.g., offset=1000, then offset=2000).data array is empty or smaller than the limit.Use the included parse_citations.py script to parse API responses and format output:
# Save and parse in one step using grep to extract JSON
curl -s "https://api.semanticscholar.org/graph/v1/paper/ARXIV:2509.16117/citations?fields=title,authors,year,url,venue&limit=1000&offset=0" \
| grep -o '{"offset".*}' \
| python3 <skill_dir>/parse_citations.py /dev/stdin --format markdown
Available formats:
markdown (default): Pretty-printed list with title, authors, year, venue, URLjson: Structured JSON arraycsv: CSV with headersARXIV: is capitalized) and ID format.-H "x-api-key: YOUR_KEY").