By name: sparql-query-files description: Executes SPARQL 1.1 queries (SELECT, ASK, CONSTRUCT, DESCRIBE) against local RDF/Turtle files in the workspace using Comunica. Use this skill whenever the user wants to query, explore, search, or analyse a knowledge graph, ontology, or vocabulary with SPARQL — for example to list all concepts, traverse class hierarchies, find related terms, inspect properties, or run custom graph queries against .ttl, .rdf, or .n3 files.
SPARQL Query Skill
⚠️ Mandatory workflow — never skip or reorder steps
Follow every step below for every query — even when the file paths or graph structure seem obvious.
Step 1 — Check existing queries
Run discover_sparql_queries. Reuse or adapt any matching SPARQL file instead of writing from scratch.
Step 2 — Discover RDF files (unless the user specified them)
Run discover_rdf_files to find all available .ttl, .rdf, .n3, .jsonld, .nt, .nq, .trig files.
Do not guess file paths — always discover first.
Step 3 — Explore graph structure
Before writing a domain-specific query, call rdf_schema_overview with the discovered files:
rdf_schema_overview(files=["path/to/file.ttl", ...])
This returns types and predicates with counts in one call — no guessing IRIs.
Step 4 — Query semantically
Traverse named individuals; use text FILTER only as a last resort.
-- ✅ Semantic traversal (preferred)
?item a ex:SomeClass ; ex:relatedTo ex:SomeOtherEntity .
-- ❌ Avoid string matching in URIs (last resort only)
FILTER(CONTAINS(LCASE(STR(?s)), "keyword"))
⚠️ Never present inferences as graph facts. Every statement must be backed by a queried triple. Mark anything else as "interpretation — not modelled in the graph".
Common patterns & prefixes
For ready-to-use query patterns and standard prefix declarations, read the reference file:
skills/sparql-query-files/reference.md