oxilite studio
A VS Code development studio for oxilite: SPARQL, Datalog and Cypher,
RDFS/OWL reasoning, SHACL and ShEx validation, over RDF stored in SQLite or Cloudflare D1.
What it does
- Project store. Every RDF file in the workspace loads into a scratch store
(
.oxilite/studio.sqlite, git-ignored), one graph per file, or as oxilite.toml says.
Editing a file reloads its graph; reasoning and validation follow.
- Languages. SPARQL (
.rq, .ru), Turtle/TriG, Datalog (.dl), Cypher (.cypher) and
ShEx (.shex): live diagnostics, completion from the store's own vocabulary (predicates by
use, classes after a, missing PREFIX lines added for you), hover, go to definition and
references across files.
- Run and explain.
Cmd/Ctrl+Enter runs a query, a rule program or a Cypher statement;
Cmd/Ctrl+Shift+E shows the SQL it compiles to. Results are a fast table, or a graph.
- Reasoning.
none, rdfs, owlql or owl2rl, plus your Datalog rules, each
materialized under its own name. The resource view marks inferred statements, and why?
shows the proof down to the lines you wrote.
- Validation. SHACL (and ShEx with a shape map) runs in the background; violations appear
in the Problems panel on the offending line, linked to the shape.
- Store Explorer. Graphs, the class hierarchy with asserted and inferred counts,
properties, files, prefixes, the ontology diagram, and a rules debugger.
- Schema registry. oxilite: Show Schema Registry draws a connection's named graphs as a
mapping graph: ontologies, shapes and ShEx graphs linked to the data graphs they apply to
(
oxl:appliesTo), imports between ontologies, inactive registrations faded and dangling
targets flagged. A per-graph table says which schemas apply to each data graph and why
(directly, through oxl:AllGraphs, or through an import). Register, remap, activate,
deactivate, unregister or drop a schema graph from the view.
- Tests.
[[test]] entries in oxilite.toml (query results, SHACL, entailments) run in
Test Explorer and in CI with oxilite check.
- Connections. Attach SQLite files, a
wrangler dev D1 database, or a remote D1 database
(read-only by default, with billed rows shown for every request). A query file can pin its
own connection with a first-line comment such as # oxilite: connection = data/prod.sqlite.
- Notebooks (
.oxnb) with SPARQL, Datalog and Cypher cells, and a kernel per connection;
the notebook remembers the one you choose.
- Agents.
oxilite mcp serves query, schema, validate and why as MCP tools; the extension
registers it, and "Copy MCP Server Config" gives the same for other agents.
Starting a project
oxilite: New Project… scaffolds a folder with a manifest, an ontology, data, a SHACL shape,
rules, a query and two tests, ready for oxilite check. It can also create a SQLite database
under db/ with a query pinned to it. oxilite: New SQLite Database… creates an empty database
and makes it the active connection. Both are on the Connections view's title bar.
A manifest
reasoning = "owl2rl"
[[graph]]
iri = "https://example.org/g/people"
files = ["data/**/*.ttl"]
[[graph]]
iri = "https://example.org/g/ontology"
files = ["ontology/*.ttl"]
role = "ontology"
applies_to = ["https://example.org/g/people"] # every graph when left out
[shapes]
files = ["shapes/*.ttl"]
[rules]
files = ["rules/*.dl"]
[[test]]
name = "people have names"
query = "tests/names.rq"
expect = "tests/names.srj"
Without a manifest the studio works on any folder: shapes, ontologies and rules are recognized
by their content and extension.
The design, decisions and milestones are in lat.md/.
Development
The engine runs in oxilite studio-server, built from the oxilite repository:
cargo build --release -p oxilite-cli # in ../oxilite
npm install && npm run build && npm test # here; the end-to-end tests use ../oxilite/target/release/oxilite
Set oxilite.server.path to the built binary, or copy it to bin/oxilite, then press F5.
| |
