Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use this skill when the user keeps paper notes inside an Obsidian project knowledge base and wants filesystem-first literature review, explicit agent-first Zotero ingestion, `Papers/` plus `Knowledge/` synthesis, collection-wide normalization, and a default literature canvas without Obsidian MCP.
.claude/skills/brycewang-stanford-obsidian-literature-workflow/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | -15% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 30% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 29% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 50% | 0% |
Handle the literature sub-workflow inside the filesystem-first Obsidian project knowledge base.
This is a supporting skill under obsidian-project-memory.
Use it when the user says things like:
Papers/”.claude/project-memory/registry.yaml, or can be imported with obsidian-project-bootstrapPapers/ inside Research/{project-slug}/$zotero-obsidian-bridge when the source corpus is in Zotero$zotero-obsidian-bridge to pull them into canonical Papers/*.md notes.Papers/ and adjacent synthesis notes using filesystem tools.$obsidian-markdown.ClaimMethodEvidenceLimitationDirect relevance to repoRelation to other papersKnowledge/ notes,Writing/ if the user asked for a review or comparison deliverable.Knowledge/, not Experiments/ or Results/:Knowledge/Literature-Overview.mdKnowledge/Method-Families.mdKnowledge/Research-Gaps.mdwhen the synthesis is stable enough to deserve canonical notes.
Maps/literature.canvas after major paper-note changes or batch note creation.Maps/literature-main.canvas only when a lightweight presentation graph is needed.Papers/ remains first-class: one durable paper note per paper whenever possibleKnowledge/ holds durable literature synthesis notesMaps/literature.canvas is the default visual graph surfaceDo not assume by default:
Concepts/Datasets/.base viewsThe literature workflow may create Maps/literature.canvas by default. Other artifacts still require explicit justification.
Load only what is needed:
references/PAPER-NOTE-SCHEMA.md - detailed paper-note frontmatter and sectionsreferences/CANVAS-WORKFLOW.md - how and when to refresh Maps/literature.canvas| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 6,363 | 3,740 | -41% | 1 | 1 | 0% | 1,110 | 1,108 | -0% | 0 | 0 | — |
case-06 | fail→pass | 10,468 | 4,387 | -58% | 1 | 1 | 0% | 1,942 | 1,660 | -15% | 0 | 0 | — |
case-02 | fail→fail | 7,758 | 4,286 | -45% | 1 | 1 | 0% | 1,347 | 1,164 | -14% | 0 | 0 | — |
case-03 | fail→fail | 12,524 | 3,182 | -75% | 1 | 1 | 0% | 2,605 | 1,050 | -60% | 0 | 0 | — |
case-04 | fail→pass | 6,673 | 3,693 | -45% | 1 | 1 | 0% | 1,242 | 1,613 | +30% | 0 | 0 | — |
case-05 | fail→pass | 13,076 | 2,851 | -78% | 1 | 1 | 0% | 2,362 | 1,418 | -40% | 0 | 0 | — |
case-07 | pass→pass | 5,043 | 3,156 | -37% | 1 | 1 | 0% | 879 | 1,437 | +63% | 0 | 0 | — |
case-08 | pass→pass | 11,355 | 10,974 | -3% | 1 | 1 | 0% | 2,086 | 2,779 | +33% | 0 | 0 | — |
case-09 | fail→pass | 5,595 | 2,288 | -59% | 1 | 1 | 0% | 972 | 1,255 | +29% | 0 | 0 | — |
case-10 | fail→pass | 10,921 | 11,376 | +4% | 1 | 1 | 0% | 1,910 | 2,867 | +50% | 0 | 0 | — |
case-11 | pass→pass | 8,572 | 4,069 | -53% | 1 | 1 | 0% | 1,500 | 1,583 | +6% | 0 | 0 | — |
case-12 | fail→pass | 11,576 | 3,745 | -68% | 1 | 1 | 0% | 1,903 | 1,575 | -17% | 0 | 0 | — |
case-13 | fail→pass | 6,356 | 1,872 | -71% | 1 | 1 | 0% | 1,402 | 1,198 | -15% | 0 | 0 | — |
case-14 | fail→pass | 6,383 | 2,242 | -65% | 1 | 1 | 0% | 1,040 | 1,273 | +22% | 0 | 0 | — |
case-15 | fail→pass | 8,571 | 4,972 | -42% | 1 | 1 | 0% | 1,766 | 1,906 | +8% | 0 | 0 | — |
case-16 | pass→pass | 6,767 | 3,605 | -47% | 1 | 1 | 0% | 1,297 | 1,520 | +17% | 0 | 0 | — |
case-17 | fail→pass | 8,142 | 1,864 | -77% | 1 | 1 | 0% | 1,448 | 1,144 | -21% | 0 | 0 | — |
case-18 | fail→pass | 6,041 | 2,539 | -58% | 1 | 1 | 0% | 955 | 1,303 | +36% | 0 | 0 | — |
case-19 | fail→fail | 6,387 | 2,422 | -62% | 1 | 1 | 0% | 1,137 | 1,313 | +15% | 0 | 0 | — |
case-20 | fail→pass | 2,985 | 10,074 | +237% | 1 | 1 | 0% | 490 | 2,773 | +466% | 0 | 0 | — |
case-21 | pass→pass | 11,488 | 10,843 | -6% | 1 | 1 | 0% | 2,290 | 2,975 | +30% | 0 | 0 | — |
case-22 | pass→pass | 7,751 | 7,732 | -0% | 1 | 1 | 0% | 1,512 | 2,316 | +53% | 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, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +55 percentage points is the difference between those two pass rates over the 20 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.