Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Access papers through interlibrary loan and document delivery services
.claude/skills/brycewang-stanford-interlibrary-loan-guide/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-02 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 33% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-01 | ✓→✓ | = Same ✓ | 28% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 61% | 0% |
A skill for accessing research papers and books through interlibrary loan (ILL) and document delivery services when your institution does not have a subscription. Covers ILL workflows, alternative free access methods, and strategies for rapid document retrieval.
Interlibrary loan is a service where your library borrows materials from another library on your behalf. Most academic libraries offer ILL free of charge to their students, faculty, and staff. Turnaround time is typically 1-7 business days for articles and 1-3 weeks for books.
Article/Chapter Request:
- You receive a digital scan (PDF) of the article
- Usually delivered to your email or ILL portal
- Turnaround: 1-5 business days
- Typically free
Book Loan:
- Physical book is shipped from another library
- Must be returned by a due date (usually 3-6 weeks)
- Turnaround: 5-15 business days
- May have a small shipping fee
Thesis/Dissertation:
- Some are available digitally via ProQuest or institutional repositories
- Others must be requested as physical loans or scans
- Turnaround varies widelypythondef prepare_ill_request(item_type: str, metadata: dict) -> dict: """ Prepare an interlibrary loan request with required information. Args: item_type: 'article', 'book', or 'chapter' metadata: Bibliographic information about the item """ required_fields = { "article": [ "article_title", "journal_title", "author", "year", "volume", "issue", "pages", "doi" ], "book": [ "title", "author", "publisher", "year", "isbn", "edition" ], "chapter": [ "chapter_title", "book_title", "author", "editor", "publisher", "year", "pages", "isbn" ] } request = {"type": item_type} fields = required_fields.get(item_type, []) for field in fields: value = metadata.get(field, "") request[field] = value if not value: request.setdefault("missing_fields", []).append(field) if request.get("missing_fields"): request["note"] = ( "Provide as many fields as possible. " "DOI or PMID alone is often sufficient for articles." ) return request
1. Verify your library does not have access
- Check library catalog and database A-Z list
- Try off-campus access via VPN or proxy
2. Gather bibliographic details
- Title, author, journal/book, year, DOI or ISBN
- The more detail you provide, the faster the request is filled
3. Submit request through your library's ILL system
- Common systems: ILLiad, Tipasa, OCLC WorldShare
- Usually accessible from your library's website under "Interlibrary Loan"
4. Wait for delivery
- Articles: PDF delivered to your email or ILL portal
- Books: Pick up at the library circulation desk
5. Return books by the due date1. Open Access repositories:
- PubMed Central (PMC) for NIH-funded biomedical research
- arXiv, bioRxiv, medRxiv for preprints
- SSRN for social science and economics working papers
- Institutional repositories (search via BASE or OpenDOAR)
2. Author contact:
- Email the corresponding author requesting a copy
- Check the author's personal or lab website for PDFs
- ResearchGate: request full text from the author
3. Legal free access tools:
- Unpaywall browser extension (finds legal OA copies)
- CORE.ac.uk (aggregates open access research)
- Google Scholar: click "PDF" links on the right side
4. Your institution:
- Try different databases (your library may have access via
a different provider)
- Ask a librarian: they know about access paths you may missWhen ILL is too slow or unavailable, commercial document delivery services can provide articles within hours:
| Service | Turnaround | Typical Cost | |---------|-----------|-------------| | British Library Document Supply | 1-2 days | Varies by country | | Reprints Desk | Same day to 48 hours | Per-article fee | | Copyright Clearance Center (Get It Now) | Minutes to hours | Per-article fee | | DeepDyve | Instant (rental model) | Monthly subscription |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | pass→pass | 12,107 | 7,475 | -38% | 1 | 1 | 0% | 1,965 | 2,520 | +28% | 0 | 0 | — |
case-02 | fail→pass | 13,206 | 8,293 | -37% | 1 | 1 | 0% | 2,244 | 2,888 | +29% | 0 | 0 | — |
case-03 | fail→pass | 9,971 | 5,846 | -41% | 1 | 1 | 0% | 1,762 | 2,339 | +33% | 0 | 0 | — |
case-04 | pass→pass | 13,667 | 12,929 | -5% | 1 | 1 | 0% | 1,998 | 3,215 | +61% | 0 | 0 | — |
case-05 | pass→pass | 13,496 | 5,655 | -58% | 1 | 1 | 0% | 2,000 | 2,077 | +4% | 0 | 0 | — |
case-06 | pass→pass | 12,592 | 7,391 | -41% | 1 | 1 | 0% | 1,817 | 2,382 | +31% | 0 | 0 | — |
case-07 | pass→pass | 12,853 | 4,453 | -65% | 1 | 1 | 0% | 1,855 | 1,948 | +5% | 0 | 0 | — |
case-08 | pass→pass | 6,755 | 4,413 | -35% | 1 | 1 | 0% | 943 | 1,947 | +106% | 0 | 0 | — |
case-09 | pass→pass | 11,914 | 8,093 | -32% | 1 | 1 | 0% | 1,887 | 2,527 | +34% | 0 | 0 | — |
case-10 | pass→pass | 7,465 | 4,590 | -39% | 1 | 1 | 0% | 1,080 | 1,892 | +75% | 0 | 0 | — |
case-11 | pass→pass | 9,576 | 4,961 | -48% | 1 | 1 | 0% | 1,339 | 1,913 | +43% | 0 | 0 | — |
case-12 | pass→pass | 13,850 | 5,904 | -57% | 1 | 1 | 0% | 1,914 | 2,084 | +9% | 0 | 0 | — |
case-13 | pass→pass | 8,475 | 5,264 | -38% | 1 | 1 | 0% | 1,198 | 2,007 | +68% | 0 | 0 | — |
case-14 | pass→pass | 16,545 | 16,671 | +1% | 1 | 1 | 0% | 2,580 | 3,779 | +46% | 0 | 0 | — |
case-15 | pass→pass | 13,389 | 10,921 | -18% | 1 | 1 | 0% | 2,051 | 2,873 | +40% | 0 | 0 | — |
case-16 | pass→pass | 12,709 | 14,195 | +12% | 1 | 1 | 0% | 1,859 | 3,165 | +70% | 0 | 0 | — |
case-17 | pass→pass | 11,968 | 10,618 | -11% | 1 | 1 | 0% | 1,934 | 2,757 | +43% | 0 | 0 | — |
case-18 | pass→pass | 16,112 | 15,078 | -6% | 1 | 1 | 0% | 2,363 | 3,393 | +44% | 0 | 0 | — |
case-19 | fail→pass | 11,134 | 7,240 | -35% | 1 | 1 | 0% | 1,770 | 2,240 | +27% | 0 | 0 | — |
case-20 | pass→pass | 5,509 | 4,804 | -13% | 1 | 1 | 0% | 1,048 | 2,241 | +114% | 0 | 0 | — |
case-21 | pass→pass | 15,557 | 16,946 | +9% | 1 | 1 | 0% | 2,295 | 3,710 | +62% | 0 | 0 | — |
case-22 | pass→pass | 25,293 | 18,707 | -26% | 1 | 1 | 0% | 2,954 | 3,902 | +32% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +14 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.