OpenLink AI Agent Skills — Knowledge Graph Infographic

Knowledge Graph Entities

Six AI Agent Skills

Each skill is modeled as a schema:SoftwareSourceCode + schema:CreativeWork entity with abstract, features, FAQ, HowTo steps, and glossary — aligned with vocabulary observed in OpenLink's URIBurner semantic descriptions.

🧠

Knowledge Graph Generator

v1.0.0

Generate comprehensive, standards-compliant Knowledge Graphs (RDF-Turtle or JSON-LD) from any file: or http(s): URL using curated schema.org-aligned prompt templates.

JSON-LDRDF-Turtleschema.orgowl:sameAsNAICS

schema:SoftwareSourceCode

schema:CreativeWork

View on ClawHub →

Key Features

Generic JSON-LD template for articles & documentation
Business & Market Analysis RDF-Turtle template
Hash-based IRI generation from @base URL
owl:sameAs DBpedia cross-referencing
Minimum 10 FAQ Q&A pairs + HowTo steps
relatedLink extraction (up to 20 URLs)
Smart quote sanitization & JSON-LD validation

Sample FAQs

Which template for business strategy content?

Business & Market Analysis (RDF-Turtle) with NAICS codes, 12 FAQs, 10 glossary terms, 7 HowTo steps.

How are DBpedia entities cross-referenced?

Via owl:sameAs — never schema:sameAs. DBpedia IRIs added to schema:about attribute.

HowTo Steps

Identify source URL (becomes @base)

Fetch page content via WebFetch

Select Generic or Business template

Populate all template placeholders

Validate, sanitize, and save to file

🗄️

Virtuoso Support Agent

v1.0.0

Technical support and database management for OpenLink Virtuoso Server with RDF Views generation, SPARQL queries, and 25 specialized MCP tools.

VirtuosoRDF ViewsMCPR2RMLSPARQL

schema:SoftwareSourceCode

schema:CreativeWork

View on ClawHub →

25 MCP Tool Categories

Entity Discovery (4): list_entity_types, list_ontologies
Database (3): schema_objects, remote_datasources, EXECUTE_SQL
RDF Views (7): FROM_TABLES, ONTOLOGY, SYNC_TO_PHYSICAL_STORE
RDF Operations (2): AUDIT_METADATA, BACKUP_METADATA
Query Execution (6): SPARQL, SPASQL, SQL, GraphQL, FacetSearch
DAV/WebDAV (3): list, get, store

Sample FAQs

What is the MCP tool naming convention?

{ServerName}:{ToolName} — e.g., Demo:execute_spasql_query or URIBurner:sparqlQuery.

What is R2RML vs Virtuoso RDF Views?

R2RML is W3C standard; Virtuoso RDF Views is OpenLink's native SQL-based approach. Both are supported.

RDF Views Workflow

Select instance (Demo or URIBurner)

Discover schema objects

Generate R2RML or RDF View mappings

Define OWL ontology from schema

Sync to physical RDF store

🌐

Wikidata Query Skill

v1.0.0

Transform natural language questions into Wikidata SPARQL queries with automatic label resolution and interactive HTML result generation.

WikidataSPARQLNL→SPARQLwikibase:labelHTML

schema:SoftwareSourceCode

schema:CreativeWork

View on ClawHub →

Key Properties Supported

P31 (instance of), P57 (director), P50 (author)
P577 (publication date), P569 (date of birth)
P19 (birthPlace), P27 (citizenship), P106 (occupation)
P1082 (population), P36 (capital)
Automatic label resolution via wikibase:label SERVICE

Sample FAQs

What is the wikibase:label service?

SERVICE block that auto-resolves Q-numbers to human-readable labels in the requested language, defaulting to English.

When to use Wikidata vs DBpedia?

Wikidata for temporal coverage, properties depth, multilingual. DBpedia for Wikipedia-aligned ontological structure.

📚

DBpedia Query Skill

v1.0.0

Transform natural language questions into DBpedia SPARQL queries with graceful error handling and beautiful HTML result pages for Wikipedia-aligned knowledge exploration.

DBpediaSPARQLWikipediadbo:HTML

schema:SoftwareSourceCode

schema:CreativeWork

View on ClawHub →

Common Property Mappings

'directed by' → dbo:director
'release date' → dbo:releaseDate or dbp:date
'budget' → dbo:budget
'born in' → dbo:birthPlace
'population' → dbo:populationTotal
'capital of' → dbo:capital

Sample FAQs

How are OPTIONAL properties handled?

OPTIONAL blocks prevent empty result sets when a property may not exist for all entities.

What LANG filter is applied?

LANG(?label) = 'en' or LANGMATCHES(LANG(?label), 'en') filters for English-language labels.

📡

RSS Feed Generator

v1.0.0

Generate valid RSS 2.0 or Atom 1.0 XML feeds from web pages that contain post lists but lack a native syndication feed.

RSS 2.0Atom 1.0FeedXMLScraping

schema:SoftwareSourceCode

schema:CreativeWork

View on ClawHub →

Extraction Heuristics

Repeated article/li/div blocks with post/entry class
h1/h2/h3 headings → item title
anchor href → item link (resolve relative URLs)
time[datetime] or text date patterns → pubDate
First paragraph (≤280 chars) → description
Author span or byline text → author field

Sample FAQs

What if the page already has a feed?

Checks for link rel='alternate' first. If found, reports URL and stops — optionally offers to proxy/mirror instead.

What is the self-hostable HTML wrapper?

An HTML page embedding the feed XML with a discovery link tag, hostable from any web server for RSS reader subscription.

⚡

OpenLink Data Twingler

v2.0.83

Execute SQL, SPARQL, SPASQL, SPARQL-FED, and GraphQL queries against live data spaces and knowledge graphs via OpenLink's OpenAPI-compliant web services.

SPARQLSPASQLSQLGraphQLRAG

schema:SoftwareSourceCode

schema:CreativeWork

View on ClawHub →

Query Language Routing

SQL: SELECT TOP 20 default; Virtuoso Demo schema
SPARQL: URIBurner endpoint; text/x-html+tr; max 20 rows
SPASQL: SPARQL wrapped in SQL FROM(...) AS alias
SPARQL-FED: triggered by named endpoint URL; SERVICE block
GraphQL: depth 10; introspection on; URIBurner endpoint

Predefined Templates

T1: Explore this Data Space
T2: Explore knowledge graph {G}
T3: Explore {G} with reasoning & inference
T4: Using endpoint {E}, explore graph {G}
T5: How to {X} (2-step HowTo pattern)

Sample FAQs

What is SPASQL?

SPARQL wrapped inside SQL: SELECT ... FROM (SPARQL SELECT ... WHERE ...) AS alias. Enables joining relational and RDF graph data.

How does it enhance LLM responses?

Routes natural language to live SPARQL/SQL/GraphQL endpoints, retrieving real-time structured data that grounds the LLM response with factual context (RAG).

Semantic Vocabulary

Knowledge Graph Topology

Terms applied across all 6 skill Knowledge Graphs, aligned with vocabulary observed from the URIBurner semantic description of the GitHub ai-agent-skills repository.

🏷️ Entity Types (rdf:type)

Type	Used For
schema:SoftwareSourceCode	Primary skill entity type (from URIBurner/GitHub)
schema:CreativeWork	Co-type for all skill entities
schema:FAQPage	FAQ section container
schema:Question	Individual FAQ question
schema:Answer	Answer to FAQ question
schema:HowTo	Step-by-step procedure
schema:HowToStep	Individual procedure step
schema:DefinedTermSet	Glossary container
schema:DefinedTerm	Individual glossary term
skos:ConceptScheme	Co-type for glossary container
schema:Organization	OpenLink Software publisher
schema:Person	Skill author (Kidehen)
schema:ItemList	Tool lists, template lists

🔧 Key Properties

Property	Value Type
schema:abstract	Language-tagged literal (from URIBurner)
schema:description	Language-tagged literal
schema:name	Language-tagged literal
schema:version	Plain literal string
schema:featureList	Array of language-tagged literals
schema:keywords	Language-tagged literal
schema:runtimePlatform	Plain literal string
schema:about	IRI array (@id values)
schema:relatedLink	IRI array (@id values)
schema:hasPart	IRI references to sections
schema:publisher	@id → Organization
schema:author	@id → Person
owl:sameAs	IRI (for DBpedia cross-refs)

🌐 Namespace Prefixes

Prefix	Namespace URI
schema:	https://schema.org/
owl:	http://www.w3.org/2002/07/owl#
rdfs:	http://www.w3.org/2000/01/rdf-schema#
skos:	http://www.w3.org/2004/02/skos/core#
dbo:	http://dbpedia.org/ontology/
dbr:	http://dbpedia.org/resource/

🔗 IRI Pattern Rules

Pattern	Example
Main skill entity	clawhub.ai/kidehen/slug
FAQ section	.../slug#faq
Question N	.../slug#q1
HowTo section	.../slug#howto
HowTo step N	.../slug#step1
Glossary	.../slug#glossary
Defined term	.../slug#term-sparql
DBpedia entity	dbpedia.org/resource/SPARQL
Publisher	openlinksw.com#OpenLinkSoftware

Endpoint	Type	Used By	Format	Limit
linkeddata.uriburner.com/sparql	SPARQL	Data Twingler, Virtuoso Agent	text/x-html+tr	20 rows
query.wikidata.org/sparql	SPARQL	Wikidata Query Skill	sparql-results+json	—
dbpedia.org/sparql	SPARQL	DBpedia Query Skill	format=json	—
linkeddata.uriburner.com/graphql	GraphQL	Data Twingler	application/json	depth 10
Demo.Demo.Customers	SQL	Data Twingler, Virtuoso Agent	tabular	TOP 20
linkeddata.uriburner.com/chat/mcp	MCP	All Skills (tier 3)	SSE / HTTP	—

Defined Terms

Cross-Skill Glossary

Key terms from the schema:DefinedTermSet entities modeled across the 6 skill Knowledge Graphs.

schema:SoftwareSourceCode

Primary entity type for all 6 skills, aligned with URIBurner's semantic description of the GitHub ai-agent-skills repository.

schema:SoftwareSourceCode

schema:abstract

Concise description property observed in URIBurner's sponge of the GitHub skills page. Used alongside schema:description for rich annotation.

schema:abstract

OPAL

OpenLink AI Layer — a Virtuoso add-on component that loosely couples LLMs with data spaces (databases, knowledge graphs, filesystems).

opal

SPASQL

Virtuoso-specific query language wrapping SPARQL inside SQL: SELECT ... FROM (SPARQL SELECT ... WHERE ...) AS alias. Enables cross-model joins.

spasql

owl:sameAs

OWL property asserting two URIs identify the same individual. Used for DBpedia cross-references in all skill KGs — never schema:sameAs.

owl:sameAs

Hash-based IRI

An IRI formed by appending a fragment (#label) to the skill's canonical URL, which serves as the JSON-LD @base. Used for all local entities.

hash-IRI

RDF Views

Virtuoso technology for generating virtual or materialized RDF triples from relational database tables via SQL-based mapping scripts.

rdf-views

SPARQL-FED

SPARQL federated query pattern using SERVICE blocks to route sub-queries to remote SPARQL endpoints while the outer query is processed locally.

sparql-fed

MCP

Model Context Protocol — a standard for exposing tools and data sources to LLMs. Used by Virtuoso Support Agent (25 tools) and all skills as tier-3 execution.

mcp

wikibase:label

Wikidata SPARQL SERVICE block that automatically resolves Q-numbers to human-readable labels in the requested language (default: English).

wikibase:label

R2RML

W3C standard mapping language for transforming relational database content into RDF. Supported alongside Virtuoso native RDF Views mappings.

r2rml

RAG

Retrieval-Augmented Generation — Data Twingler's core pattern: live query execution against SPARQL/SQL endpoints augments LLM responses with factual data.

rag

Frequently Asked Questions

Selected FAQs Across Skills

🧠 KG Generator

Generic template (JSON-LD) — optimized for articles, blogs, and documentation using schema.org vocabulary with hash-based IRIs.

Generic template: minimum 10 Q&A pairs. Business & Market Analysis template: exactly 12 questions, 10 glossary terms, 7 HowTo steps.

All IRI-valued attributes must use @id in JSON-LD. Plain string literals are never assigned to IRI-only properties like owl:sameAs.

🗄️ Virtuoso Support Agent

After RDF Views are created as virtual mappings, RDFVIEW_SYNC_TO_PHYSICAL_STORE materializes them into persistent triples in Virtuoso's quad store for faster SPARQL performance.

Demo (test/sample data with Northwind at demo.openlinksw.com) and URIBurner (production at linkeddata.uriburner.com). Prefix all tools with {ServerName}:.

Virtuoso's SPARQL-within-SQL syntax: SELECT ... FROM (SPARQL SELECT ... WHERE ...) AS alias. Enables joining relational and RDF graph data in one query.

🌐 Wikidata Query Skill

Properties use wdt:P### (e.g., wdt:P57 for director). Entities use wd:Q### (e.g., wd:Q51566 for Spike Lee). Label resolution via wikibase:label SERVICE.

Wikidata: better temporal coverage, more properties, multilingual. DBpedia: Wikipedia-aligned entities, strong ontological structure (dbo: vocabulary).

⚡ Data Twingler

When the user explicitly names a SPARQL endpoint URL in the prompt. Named endpoint becomes a SERVICE block; default URIBurner endpoint processes the outer query.

T1: Explore Data Space · T2: Explore named KG · T3: KG with reasoning · T4: SPARQL-FED with endpoint · T5: HowTo 2-step. Loaded from references/query-templates.md.

Routes natural language to live query endpoints (SPARQL/SQL/GraphQL), retrieving real-time structured data that grounds the LLM response — implementing a RAG pattern.

Six Skills.
One Semantic Web.

Six AI Agent Skills

Knowledge Graph Topology

Multi-Protocol Execution Flow

Service Endpoints Across Skills

Cross-Skill Glossary

Selected FAQs Across Skills

Resources & References