Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Search computer science publications, authors, and venues via DBLP
.claude/skills/brycewang-stanford-dblp-api/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 45% | 0% |
| case-09 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 10% | 0% |
| case-17 | ✗→✓ | ▲ Improved | 99% | 0% |
| case-18 | ✗→✓ | ▲ Improved | 26% | 0% |
DBLP (Digital Bibliography and Library Project) is the most comprehensive open bibliographic database for computer science. Maintained by Schloss Dagstuhl - Leibniz Center for Informatics, DBLP indexes over 6.5 million publications from thousands of journals, conference proceedings, and informal publication venues in computer science and related fields.
DBLP is widely regarded as the authoritative source for computer science bibliographic data. It is used by researchers for tracking publication records, by hiring committees for evaluating academic productivity, and by bibliometric tools for analyzing the computer science research landscape. Each author in DBLP has a unique profile page with a complete, disambiguated publication list.
The API is entirely free and open, requiring no authentication. There are no published rate limits, though users are expected to be respectful and avoid excessive load on the servers. All data is available under the ODC-BY license.
No authentication required. The DBLP API is fully open and free to use. No API key, registration, or email is needed. Users should include a descriptive User-Agent header and avoid sending excessive concurrent requests.
GET https://dblp.org/search/publ/api| Param | Type | Required | Description | |-------|------|----------|-------------| | q | string | Yes | Search query (supports author:, venue:, year: prefixes) | | format | string | No | Response format: json, xml (default: xml) | | h | integer | No | Number of results (default: 30, max: 1000) | | f | integer | No | Starting offset for pagination (default: 0) | | c | integer | No | Maximum completions in auto-suggest mode |
bash curl "https://dblp.org/search/publ/api?q=graph+neural+networks&format=json&h=10&f=0"
result.hits containing @total count and hit array. Each hit has info with title, authors.author, venue, year, type, doi, url, and ee (electronic edition link).GET https://dblp.org/search/author/api| Param | Type | Required | Description | |-------|------|----------|-------------| | q | string | Yes | Author name query | | format | string | No | Response format: json, xml | | h | integer | No | Number of results (default: 30) | | f | integer | No | Starting offset |
bash curl "https://dblp.org/search/author/api?q=Yann+LeCun&format=json&h=5"
author name, url (DBLP profile URL), and optional notes with affiliation information.GET https://dblp.org/search/venue/api| Param | Type | Required | Description | |-------|------|----------|-------------| | q | string | Yes | Venue name query | | format | string | No | Response format: json, xml | | h | integer | No | Number of results | | f | integer | No | Starting offset |
bash curl "https://dblp.org/search/venue/api?q=NeurIPS&format=json&h=5"
GET https://dblp.org/pid/{pid}.json| Param | Type | Required | Description | |-------|------|----------|-------------| | pid | string | Yes | DBLP person ID (e.g., h/GeoffreyEHinton) from author search results |
bash curl "https://dblp.org/pid/h/GeoffreyEHinton.json"
No published rate limits. DBLP does not enforce strict API quotas. However, the service is provided by a non-profit institution with limited resources. Best practices include limiting requests to a reasonable rate (1-2 per second), using caching for repeated queries, and downloading the DBLP XML dump for large-scale analyses instead of querying the API repeatedly.
Get a complete, chronological list of publications for a researcher:
bash# Step 1: Find the author's DBLP PID curl "https://dblp.org/search/author/api?q=Jure+Leskovec&format=json&h=1" # Step 2: Fetch their full profile curl "https://dblp.org/pid/l/JureLeskovec.json"
Search for publications at a specific venue in a given year:
bashcurl "https://dblp.org/search/publ/api?q=venue:ICML+year:2024&format=json&h=50"
Use DBLP records to find DOIs and link to other metadata sources:
bashcurl "https://dblp.org/search/publ/api?q=attention+is+all+you+need&format=json&h=1" | jq '.result.hits.hit[0].info.doi'
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 13,908 | 21,217 | +53% | 1 | 1 | 0% | 2,762 | 4,002 | +45% | 0 | 0 | — |
case-02 | pass→pass | 6,666 | 2,775 | -58% | 1 | 1 | 0% | 1,072 | 1,906 | +78% | 0 | 0 | — |
case-03 | pass→pass | 6,683 | 3,804 | -43% | 1 | 1 | 0% | 1,187 | 2,107 | +78% | 0 | 0 | — |
case-04 | pass→pass | 4,373 | 2,597 | -41% | 1 | 1 | 0% | 685 | 1,847 | +170% | 0 | 0 | — |
case-05 | pass→pass | 9,224 | 4,562 | -51% | 1 | 1 | 0% | 1,576 | 2,347 | +49% | 0 | 0 | — |
case-06 | pass→pass | 6,957 | 2,643 | -62% | 1 | 1 | 0% | 1,176 | 1,924 | +64% | 0 | 0 | — |
case-07 | pass→pass | 6,016 | 2,690 | -55% | 1 | 1 | 0% | 1,133 | 1,879 | +66% | 0 | 0 | — |
case-08 | pass→pass | 5,986 | 2,156 | -64% | 1 | 1 | 0% | 1,065 | 1,835 | +72% | 0 | 0 | — |
case-09 | fail→pass | 16,173 | 6,327 | -61% | 1 | 1 | 0% | 2,799 | 2,384 | -15% | 0 | 0 | — |
case-10 | pass→pass | 8,820 | 6,127 | -31% | 1 | 1 | 0% | 1,515 | 2,447 | +62% | 0 | 0 | — |
case-11 | pass→pass | 7,670 | 4,574 | -40% | 1 | 1 | 0% | 1,377 | 2,213 | +61% | 0 | 0 | — |
case-12 | pass→pass | 8,746 | 4,554 | -48% | 1 | 1 | 0% | 1,478 | 2,132 | +44% | 0 | 0 | — |
case-13 | fail→pass | 11,642 | 2,899 | -75% | 1 | 1 | 0% | 1,742 | 1,922 | +10% | 0 | 0 | — |
case-14 | pass→pass | 14,036 | 8,829 | -37% | 1 | 1 | 0% | 2,011 | 2,765 | +37% | 0 | 0 | — |
case-15 | pass→pass | 5,171 | 3,902 | -25% | 1 | 1 | 0% | 894 | 2,018 | +126% | 0 | 0 | — |
case-16 | pass→pass | 4,111 | 2,115 | -49% | 1 | 1 | 0% | 655 | 1,745 | +166% | 0 | 0 | — |
case-17 | fail→pass | 5,678 | 1,844 | -68% | 1 | 1 | 0% | 839 | 1,667 | +99% | 0 | 0 | — |
case-18 | fail→pass | 9,758 | 4,332 | -56% | 1 | 1 | 0% | 1,746 | 2,197 | +26% | 0 | 0 | — |
case-19 | pass→pass | 9,472 | 8,746 | -8% | 1 | 1 | 0% | 1,635 | 2,681 | +64% | 0 | 0 | — |
case-20 | pass→pass | 6,489 | 3,480 | -46% | 1 | 1 | 0% | 1,009 | 1,989 | +97% | 0 | 0 | — |
case-21 | pass→pass | 14,945 | 11,977 | -20% | 1 | 1 | 0% | 2,798 | 3,433 | +23% | 0 | 0 | — |
case-22 | pass→pass | 7,161 | 5,241 | -27% | 1 | 1 | 0% | 1,197 | 2,310 | +93% | 0 | 0 | — |
case-23 | pass→pass | 10,443 | 5,958 | -43% | 1 | 1 | 0% | 1,661 | 2,359 | +42% | 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. 23 cases were attempted. The headline lift of +22 percentage points is the difference between those two pass rates over the 23 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.