Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Insert citations and notes from Zotero into Obsidian knowledge bases
.claude/skills/brycewang-stanford-obsidian-zotero-guide/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 26% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-15 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-19 | ✗→✓ | ▲ Improved | 1% | 0% |
A skill for integrating Zotero reference management with Obsidian note-taking to create a seamless academic knowledge management workflow. Based on the obsidian-zotero-integration plugin (2K stars), this skill guides the agent through setting up, configuring, and optimizing the bridge between your reference library and your personal knowledge base.
Academic researchers accumulate hundreds or thousands of papers over a career, each containing insights that need to be connected, synthesized, and retrieved when writing new work. Zotero excels at organizing references and PDFs, while Obsidian excels at creating interlinked notes and building knowledge graphs. This integration bridges the two, allowing annotations made in Zotero to flow into Obsidian as linked notes, and enabling citation insertion directly from the Obsidian editor.
The result is a unified research workflow where reading, annotating, note-taking, and writing happen in a connected ecosystem rather than in isolated tools.
Prerequisites
references/ or literature/)Installation Steps
authYear or authYearTitle)Template Configuration
The primary citation workflow during writing:
Inserting Citations
Managing Literature Notes
Building Connections
One of the most powerful features is extracting PDF annotations:
Highlight Extraction
Comment Extraction
Image Extraction
Systematic Review Support
Collaborative Research
Writing Integration
This skill enhances the Research-Claw reference management capabilities:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 19,692 | 21,473 | +9% | 1 | 1 | 0% | 3,868 | 4,881 | +26% | 0 | 0 | — |
case-02 | fail→pass | 29,080 | 23,488 | -19% | 1 | 1 | 0% | 4,052 | 4,894 | +21% | 0 | 0 | — |
case-03 | pass→pass | 15,762 | 15,745 | -0% | 1 | 1 | 0% | 2,347 | 3,991 | +70% | 0 | 0 | — |
case-04 | pass→pass | 20,040 | 12,818 | -36% | 1 | 1 | 0% | 3,252 | 3,626 | +12% | 0 | 0 | — |
case-05 | pass→pass | 10,068 | 8,826 | -12% | 1 | 1 | 0% | 1,786 | 2,803 | +57% | 0 | 0 | — |
case-06 | pass→pass | 5,901 | 2,323 | -61% | 1 | 1 | 0% | 814 | 1,562 | +92% | 0 | 0 | — |
case-07 | pass→pass | 16,810 | 17,271 | +3% | 1 | 1 | 0% | 2,405 | 3,686 | +53% | 0 | 0 | — |
case-08 | fail→fail | 13,091 | 11,164 | -15% | 1 | 1 | 0% | 2,202 | 3,011 | +37% | 0 | 0 | — |
case-09 | fail→pass | 14,189 | 14,299 | +1% | 1 | 1 | 0% | 2,019 | 3,318 | +64% | 0 | 0 | — |
case-10 | pass→pass | 11,567 | 7,165 | -38% | 1 | 1 | 0% | 1,842 | 2,428 | +32% | 0 | 0 | — |
case-11 | fail→fail | 10,951 | 9,348 | -15% | 1 | 1 | 0% | 1,792 | 2,798 | +56% | 0 | 0 | — |
case-12 | fail→fail | 10,046 | 7,413 | -26% | 1 | 1 | 0% | 1,642 | 2,521 | +54% | 0 | 0 | — |
case-13 | pass→pass | 15,584 | 10,640 | -32% | 1 | 1 | 0% | 2,119 | 2,850 | +34% | 0 | 0 | — |
case-14 | pass→pass | 12,055 | 12,986 | +8% | 1 | 1 | 0% | 1,991 | 3,069 | +54% | 0 | 0 | — |
case-15 | fail→pass | 13,026 | 11,071 | -15% | 1 | 1 | 0% | 2,119 | 2,802 | +32% | 0 | 0 | — |
case-16 | pass→pass | 14,295 | 11,523 | -19% | 1 | 1 | 0% | 2,142 | 3,080 | +44% | 0 | 0 | — |
case-17 | pass→pass | 19,306 | 4,644 | -76% | 1 | 1 | 0% | 2,189 | 1,876 | -14% | 0 | 0 | — |
case-18 | pass→pass | 14,385 | 12,304 | -14% | 1 | 1 | 0% | 2,374 | 2,985 | +26% | 0 | 0 | — |
case-19 | fail→pass | 11,136 | 2,687 | -76% | 1 | 1 | 0% | 1,653 | 1,671 | +1% | 0 | 0 | — |
case-20 | pass→pass | 18,956 | 58,543 | +209% | 1 | 1 | 0% | 2,698 | 4,234 | +57% | 0 | 0 | — |
case-21 | pass→pass | 17,013 | 10,219 | -40% | 1 | 1 | 0% | 2,585 | 2,627 | +2% | 0 | 0 | — |
case-22 | fail→pass | 14,701 | 6,968 | -53% | 1 | 1 | 0% | 2,033 | 2,228 | +10% | 0 | 0 | — |
case-23 | pass→pass | 17,825 | 16,979 | -5% | 1 | 1 | 0% | 2,948 | 4,117 | +40% | 0 | 0 | — |
case-24 | pass→pass | 15,797 | 10,863 | -31% | 1 | 1 | 0% | 2,291 | 2,938 | +28% | 0 | 0 | — |
case-25 | pass→pass | 15,489 | 6,692 | -57% | 1 | 1 | 0% | 2,091 | 2,353 | +13% | 0 | 0 | — |
case-26 | pass→pass | 20,970 | 21,825 | +4% | 1 | 1 | 0% | 2,891 | 4,059 | +40% | 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. 26 cases were attempted. The headline lift of +23 percentage points is the difference between those two pass rates over the 26 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.