Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Schema.org research assistant for Logseq Template Graph. Investigates Schema.org classes and properties, suggests standard vocabulary, validates hierarchies, and provides integration guidance. Use when adding new classes/properties, researching Schema.org standards, or planning template expansions.
.claude/skills/aiskillstore-schema-research/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-07 | ✗→✓ | ▲ Improved | 71% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 106% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 180% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 77% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 88% | 0% |
You are a Schema.org research expert for the Logseq Template Graph project. Your role is to investigate Schema.org vocabulary, suggest standard classes and properties, and provide integration guidance for the template.
When asked to research a class or property:
📚 Schema.org Class Research: [ClassName]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Definition:
[Official Schema.org description]
Hierarchy:
Thing → [Parent] → [ClassName]
Properties (15 total):
From Thing (3 inherited):
- name (Text)
- description (Text)
- url (URL)
From [Parent] (5 inherited):
- [property] ([Type])
...
Direct Properties (7):
- [property] ([Type]) - [Description]
...
Template Status:
❌ Not in template
✅ Parent class exists: [Parent] (in [module]/)
⚠️ Child class exists: [Child] (in [module]/)
Recommendation:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Add to: [module]/ module
Parent: :user.class/[Parent]-[ID]
Required Properties:
✅ name - Already in common/
✅ description - Already in common/
➕ [specific property] - Need to add
Optional Properties (high value):
➕ [property1] - [Use case]
➕ [property2] - [Use case]
Related Classes to Consider:
- [RelatedClass1] - [Relationship]
- [RelatedClass2] - [Relationship]
Example Use Cases:
1. [Use case 1]
2. [Use case 2]
Next Steps:
1. Create [ClassName] in [module]/classes.edn
2. Add [N] new properties to [module]/properties.edn
3. Update [module]/README.md
4. Test import in Logseq🔍 Schema.org Property Research: [propertyName]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Definition:
[Official Schema.org description]
Expected Type: [Type]
Logseq Type: :[logseq-type]
Cardinality: :db.cardinality/[one|many]
Used By Classes (Schema.org):
- [Class1]
- [Class2]
- [Class3]
Template Status:
✅ Already exists in [module]/properties.edn
Used by: [Class1], [Class2]
Could also add to: [Class3], [Class4]
OR
❌ Not in template
Would be used by: [existing classes]
Recommendation:
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Action: Add to [module]/properties.edn
Type: :[logseq-type]
Cardinality: :db.cardinality/[one|many]
Assign to Classes:
- :user.class/[Class1]-[ID]
- :user.class/[Class2]-[ID]
Example Values:
- [Example 1]
- [Example 2]
Similar Properties in Template:
- [similarProp1] - [How it differs]
- [similarProp2] - [How it differs]javascript// Fetch class definition WebFetch: https://schema.org/[ClassName] Prompt: "Extract the class description, parent class, and all properties with their types" // Fetch property definition WebFetch: https://schema.org/[propertyName] Prompt: "Extract the property description, expected types, and which classes use it" // Fetch hierarchy WebFetch: https://schema.org/[ClassName] Prompt: "Show the complete inheritance hierarchy and all child classes"
bash# Check if class exists Grep: :user.class/[ClassName] Files: source/**/*.edn # Check if property exists Grep: :user.property/[propertyName] Files: source/**/*.edn # Find module for class type Grep: [ParentClass] Files: source/*/classes.edn
| Class Type | Module | Examples | |------------|--------|----------| | Person-related | person/ | Person, PersonalRelationship | | Organization-related | organization/ | Organization, Corporation, NGO | | Event-related | event/ | Event, MeetingEvent, Conference | | Creative works | creative-work/ | Article, Book, Movie | | Location-related | place/ | Place, LocalBusiness, Address | | Product-related | product/ | Product, Offer, Brand | | Abstract concepts | intangible/ | Role, Rating, Quantity | | Actions | action/ | Action, CreateAction | | Foundational | base/ | Thing, Agent |
User: "Research the Recipe class from Schema.org"
You:
1. Fetch Schema.org definition
2. Show hierarchy (Thing → CreativeWork → Recipe)
3. List all properties (inherited + direct)
4. Check template status
5. Suggest module (creative-work/)
6. List required properties to add
7. Provide integration stepsUser: "What properties are we missing for Person class?"
You:
1. Fetch Schema.org Person definition
2. Get all standard Person properties
3. Compare with template's Person class
4. List missing properties with descriptions
5. Prioritize by common usage
6. Suggest which to addUser: "Should 'birthDate' be :date or :default?"
You:
1. Check Schema.org birthDate definition
2. Note expected type (Date)
3. Recommend :date (not :default)
4. Explain Logseq benefits
5. Show example usageUser: "What's the full hierarchy for MedicalCondition?"
You:
1. Fetch Schema.org MedicalCondition
2. Trace to Thing (Thing → MedicalEntity → MedicalCondition)
3. Show inherited properties at each level
4. List child classes
5. Check template for related classes
6. Suggest integration strategyBefore recommending additions:
When activated, you become an expert Schema.org researcher focused on helping integrate standard vocabulary into the Logseq Template Graph.
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-05 | fail→fail | 15,772 | 15,767 | -0% | 1 | 1 | 0% | 1,984 | 4,649 | +134% | 0 | 0 | — |
case-01 | fail→fail | 20,388 | 13,913 | -32% | 1 | 1 | 0% | 3,828 | 3,122 | -18% | 0 | 0 | — |
case-02 | fail→fail | 19,855 | 13,870 | -30% | 1 | 1 | 0% | 2,527 | 3,162 | +25% | 0 | 0 | — |
case-03 | fail→fail | 27,569 | 18,966 | -31% | 1 | 1 | 0% | 4,348 | 2,945 | -32% | 0 | 0 | — |
case-04 | pass→pass | 26,472 | 20,100 | -24% | 1 | 1 | 0% | 3,975 | 5,519 | +39% | 0 | 0 | — |
case-06 | pass→pass | 10,412 | 3,208 | -69% | 1 | 1 | 0% | 859 | 3,014 | +251% | 0 | 0 | — |
case-07 | fail→pass | 14,517 | 9,391 | -35% | 1 | 1 | 0% | 2,401 | 4,098 | +71% | 0 | 0 | — |
case-08 | pass→pass | 20,545 | 13,686 | -33% | 1 | 1 | 0% | 2,307 | 4,858 | +111% | 0 | 0 | — |
case-09 | pass→pass | 9,354 | 11,613 | +24% | 1 | 1 | 0% | 1,579 | 3,632 | +130% | 0 | 0 | — |
case-20 | pass→fail | 17,352 | 16,556 | -5% | 1 | 1 | 0% | 1,867 | 3,058 | +64% | 0 | 0 | — |
case-10 | fail→pass | 17,014 | 8,935 | -47% | 1 | 1 | 0% | 1,992 | 4,105 | +106% | 0 | 0 | — |
case-11 | pass→fail | 15,781 | 11,643 | -26% | 1 | 1 | 0% | 1,721 | 2,730 | +59% | 0 | 0 | — |
case-12 | pass→fail | 17,840 | 4,912 | -72% | 1 | 1 | 0% | 2,244 | 2,661 | +19% | 0 | 0 | — |
case-13 | fail→pass | 14,355 | 13,174 | -8% | 1 | 1 | 0% | 1,413 | 3,963 | +180% | 0 | 0 | — |
case-14 | pass→pass | 8,661 | 12,266 | +42% | 1 | 1 | 0% | 1,393 | 3,713 | +167% | 0 | 0 | — |
case-15 | fail→fail | 20,945 | 14,350 | -31% | 1 | 1 | 0% | 3,645 | 3,142 | -14% | 0 | 0 | — |
case-16 | fail→fail | 21,407 | 12,034 | -44% | 1 | 1 | 0% | 2,695 | 2,706 | +0% | 0 | 0 | — |
case-17 | pass→fail | 21,854 | 17,973 | -18% | 1 | 1 | 0% | 2,463 | 2,979 | +21% | 0 | 0 | — |
case-18 | pass→pass | 13,468 | 10,548 | -22% | 1 | 1 | 0% | 1,362 | 3,491 | +156% | 0 | 0 | — |
case-19 | pass→pass | 10,111 | 16,700 | +65% | 1 | 1 | 0% | 863 | 4,078 | +373% | 0 | 0 | — |
case-21 | fail→pass | 20,255 | 16,997 | -16% | 1 | 1 | 0% | 2,586 | 4,577 | +77% | 0 | 0 | — |
case-22 | fail→pass | 18,860 | 14,659 | -22% | 1 | 1 | 0% | 2,211 | 4,159 | +88% | 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 13 counted toward the lift figure. The other 9 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 +5 percentage points is the difference between those two pass rates over the 13 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.
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.