yurtle-rdflib

RDFlib plugin for Yurtle format - Markdown files with Turtle/YAML frontmatter that form a queryable knowledge graph.

What is Yurtle?

Yurtle (YAML/RDF + Turtle) is a file format that combines:

• Markdown content for human-readable documentation
• Turtle or YAML frontmatter for machine-readable RDF triples
• Yurtle blocks for inline structured data anywhere in the document

This enables every .md file to be both a document AND a node in a knowledge graph.

---
yurtle: v1.3
id: nautical-project/ship
type: asset
title: Clipper Windchaser
relates-to:
  - nautical-project/voyage
  - nautical-project/crew
nugget: 1852 three-masted clipper - fastest trade-wind runner in the fleet
---

# Windchaser

Built in Aberdeen, copper-sheathed, Baltimore clipper lines.

` ` `yurtle
ship:
  name: Windchaser
  built: 1852
  length: 62m
  captain: crew-captain-reed
` ` `

She has outrun typhoons and carried tea from Canton in 79 days.

(Remove spaces in fence markers above - shown for display)

Why Integrate Yurtle with RDFlib?

Yurtle files are human-readable and LLM-friendly on their own. So why add RDFlib?

1. SPARQL Queries Across Your Entire Workspace

Without RDFlib, finding "all lab tests related to diabetes" means writing custom code to parse each file. With RDFlib:

graph.query("""
    SELECT ?test ?loinc WHERE {
        ?test med:clinicalUse ?use .
        FILTER(CONTAINS(?use, "diabetes"))
    }
""")

One query. Any complexity. Aggregations, joins, filters, graph traversal — all built-in.

2. Interoperability with Standard Vocabularies

RDFlib connects your Yurtle files to the wider semantic web:

• LOINC for laboratory tests
• SNOMED-CT for medical terminology
• Schema.org for web content
• Dublin Core for document metadata
• Any RDF vocabulary you need

3. Bidirectional Sync: Files ↔ Graph

Edit files in your favorite editor (Obsidian, VSCode, vim) — the graph updates. Modify the graph programmatically — files update automatically.

kb = create_live_graph("workspace/", auto_flush=True)
kb.add((subject, predicate, object))  # File written immediately

4. Provenance Tracking

Every triple knows where it came from:

# Find which file defines a concept
graph.query("""
    SELECT ?file WHERE {
        ?concept yurtle:title "Diabetes" .
        ?concept prov:definedIn ?file .
    }
""")

5. LLM + Structured Queries = Best of Both

• LLMs excel at understanding prose and answering open-ended questions
• SPARQL excels at precise, repeatable, complex queries

Use both: LLMs read the markdown content, SPARQL queries the structured data.

6. No Database Required

Your knowledge graph is just markdown files in a folder:

• Version control with Git
• Edit anywhere
• No server to maintain
• Works offline

Installation

pip install yurtle-rdflib

Quick Start

Parse a Yurtle file

from rdflib import Graph
import yurtle_rdflib  # Registers the plugin

# Parse a single file
graph = Graph()
graph.parse("document.md", format="yurtle")

# Query with SPARQL
results = graph.query("""
    SELECT ?title ?status WHERE {
        ?task yurtle:title ?title ;
              pm:status ?status .
    }
""")

Load an entire workspace

import yurtle_rdflib

# Load all .md files in a directory
graph = yurtle_rdflib.load_workspace("my-project/")

# Query across all documents
results = graph.query("""
    SELECT ?doc ?title WHERE {
        ?doc yurtle:title ?title .
    }
""")

Serialize to Yurtle format

from rdflib import Graph, URIRef, Literal
import yurtle_rdflib

graph = Graph()
graph.add((
    URIRef("urn:task:T-001"),
    yurtle_rdflib.YURTLE.title,
    Literal("My Task")
))

# Serialize with markdown content
graph.serialize(
    "output.md",
    format="yurtle",
    markdown_content="# My Task\n\nDescription here..."
)

Live bidirectional sync

import yurtle_rdflib

# Create a graph backed by the filesystem
graph = yurtle_rdflib.create_live_graph("workspace/", auto_flush=True)

# Changes persist immediately to files
graph.add((subject, predicate, object))

# File changes sync automatically
graph.store.sync()

Features

Parser

• Turtle frontmatter: Native RDF parsing
• YAML frontmatter: Automatic conversion to RDF triples
• Provenance tracking: Each triple knows its source file
• Works with rdflib: graph.parse("file.md", format="yurtle")

Serializer

• Round-trip safe: Preserves markdown content
• Clean output: Filters internal provenance triples
• Standard prefixes: Auto-binds common namespaces
• Works with rdflib: graph.serialize("file.md", format="yurtle")

YurtleStore

• Bidirectional sync: Graph ↔ Filesystem
• Hash-based change detection: Efficient incremental sync
• Auto-flush mode: Immediate persistence
• SPARQL queries: Query files as a graph

Standard Namespaces

from yurtle_rdflib import YURTLE, PM, BEING, VOYAGE, KNOWLEDGE, PROVENANCE

# Use in your graphs
graph.add((subject, YURTLE.title, Literal("My Document")))
graph.add((task, PM.status, Literal("completed")))

Prefix	Namespace	Purpose
yurtle	https://yurtle.dev/schema/	Core document properties
pm	https://yurtle.dev/pm/	Project management
being	https://yurtle.dev/being/	Agent/being properties
voyage	https://yurtle.dev/voyage/	Journey/process tracking
knowledge	https://yurtle.dev/knowledge/	Learning/knowledge
prov	https://yurtle.dev/provenance/	Source file tracking

API Reference

Functions

Function	Description
load_workspace(path)	Load all Yurtle files into a unified graph
save_workspace(graph, path)	Save graph back to workspace files
create_live_graph(path)	Create store-backed graph with live sync
parse_file(path)	Parse a single Yurtle file
serialize_file(graph, path)	Serialize graph to Yurtle file
verify_plugins()	Check plugin registration status

Classes

Class	Description
YurtleParser	Core parser for Yurtle documents
YurtleWriter	Core writer for Yurtle documents
YurtleDocument	Parsed document with graph + content
YurtleStore	RDFlib Store for bidirectional sync
YurtleRDFlibParser	RDFlib Parser plugin
YurtleRDFlibSerializer	RDFlib Serializer plugin

Examples

See the examples/ directory for complete workspaces:

• nautical-project/ - Sailing voyage knowledge graph (from Yurtle spec)
• lab-tests/ - Medical laboratory tests with LOINC codes
• QUERIES.md - 20+ SPARQL query examples

Load the Nautical Project

import yurtle_rdflib

# Load the example workspace
graph = yurtle_rdflib.load_workspace("examples/nautical-project/")

# Find all crew members
results = graph.query("""
    PREFIX yurtle: <https://yurtle.dev/schema/>
    SELECT ?person ?name ?role WHERE {
        ?person yurtle:type "person" ;
                yurtle:title ?name .
        OPTIONAL { ?person yurtle:role ?role }
    }
""")

for row in results:
    print(f"{row.name}: {row.role or 'crew'}")

Query Laboratory Tests

import yurtle_rdflib

# Load lab tests
graph = yurtle_rdflib.load_workspace("examples/lab-tests/")

# Find diabetes-related tests
results = graph.query("""
    PREFIX lab: <https://yurtle.dev/lab/>
    PREFIX med: <https://yurtle.dev/medical/>
    PREFIX yurtle: <https://yurtle.dev/schema/>

    SELECT ?test ?title ?loinc WHERE {
        ?test a lab:LaboratoryTest ;
              yurtle:title ?title ;
              lab:loincCode ?loinc ;
              med:clinicalUse ?use .
        FILTER(CONTAINS(LCASE(?use), "diabetes"))
    }
""")

for row in results:
    print(f"{row.title} (LOINC {row.loinc})")

Advanced: Graph Traversal

# Find all items related to the voyage (2 hops)
results = graph.query("""
    PREFIX yurtle: <https://yurtle.dev/schema/>

    SELECT ?item ?title ?path WHERE {
        ?voyage yurtle:title "Voyage to the Sapphire Isles" .
        {
            ?voyage yurtle:relatesTo ?item .
            BIND("direct" as ?path)
        }
        UNION
        {
            ?voyage yurtle:relatesTo ?middle .
            ?middle yurtle:relatesTo ?item .
            BIND("via " + STR(?middle) as ?path)
        }
        ?item yurtle:title ?title .
    }
""")

Advanced: Aggregation

# Count documents by type
results = graph.query("""
    PREFIX yurtle: <https://yurtle.dev/schema/>

    SELECT ?type (COUNT(?doc) as ?count) WHERE {
        ?doc yurtle:type ?type .
    }
    GROUP BY ?type
    ORDER BY DESC(?count)
""")

for row in results:
    print(f"{row.type}: {row.count}")

Live Sync with Filesystem

import yurtle_rdflib
from rdflib import URIRef, Literal

# Create live-synced knowledge base
kb = yurtle_rdflib.create_live_graph("my-workspace/", auto_flush=True)

# Add knowledge (persists immediately to files)
kb.add((
    URIRef("urn:concept:python"),
    yurtle_rdflib.KNOWLEDGE.relatedTo,
    URIRef("urn:concept:programming")
))

# Changes in files sync back automatically
kb.store.sync()

Yurtle Format Specification

See the Yurtle Specification for the full format documentation.

Supported Frontmatter Types

Turtle (recommended):

---
@prefix yurtle: <https://yurtle.dev/schema/> .

<urn:doc:example> a yurtle:Document ;
    yurtle:title "Example" .
---

YAML (auto-converted):

---
id: example
title: Example
type: document
tags: [example, demo]
---

Development

# Clone the repo
git clone https://github.com/hankh95/yurtle-rdflib.git
cd yurtle-rdflib

# Install with dev dependencies
pip install -e ".[dev]"

# Run tests
pytest

# Run with coverage
pytest --cov=yurtle_rdflib

# Format code
black src tests
ruff check src tests

Changelog

v0.2.0 (2026-03-01)

Fenced block parsing and robustness.

Added

• Fenced turtle/yurtle block parsing — Parse ```turtle and ```yurtle fenced blocks within Yurtle markdown files. Triples from fenced blocks are merged into the same graph as frontmatter triples (EXP-1024)

Fixed

• Prefix inheritance — Fenced blocks inherit @prefix declarations from the YAML/Turtle frontmatter, eliminating "Prefix not bound" warnings (PR #3)
• YAML frontmatter skip — Parser correctly skips pure YAML frontmatter (no Turtle) without emitting parse errors
• Git conflict marker skip — Files containing <<<<<<< merge conflict markers are silently skipped instead of crashing

v0.1.0 (2026-02-22)

• Initial release: RDFlib plugin for parsing and serializing Yurtle files
• SPARQL-queryable graphs from Yurtle markdown with Turtle frontmatter
• Registered as yurtle format with RDFlib's plugin system

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE for details.

Related Projects

• Yurtle Specification - The Yurtle format spec
• Yurtle Kanban - File-based kanban on git
• RDFlib - Python RDF library