MCP¶

jena-mcp exposes the same capability three ways: as MCP tools an agent calls, as a Python API (JenaApi) you import, and as a CLI. The complete tool surface is summarized in Overview.

As an MCP server¶

Once deployed, the server registers three action-dispatch tools that cover the SPARQL Protocol, the Graph Store Protocol, and Fuseki administration:

Tool	Actions
`jena_sparql`	`query` (SELECT/ASK/CONSTRUCT/DESCRIBE), `update` (INSERT/DELETE/LOAD/CLEAR)
`jena_graph`	`get`, `put` (replace), `post` (merge), `delete` — via the Graph Store Protocol
`jena_admin`	`ping`, `server_info`, `stats`, `metrics`, `list_datasets`, `dataset_info`, `create_dataset`, `delete_dataset`, `set_dataset_state`, `list_tasks`, `task_info`, `backup`, `compact`

Example agent prompts that map onto these tools:

"List the datasets on the Fuseki server" → jena_admin (list_datasets)
"Run this SELECT query against dataset ds" → jena_sparql (query)
"Load this Turtle into the default graph of ds" → jena_graph (post)

As a Python API¶

JenaApi (exported as Api) is a requests-based facade over a Fuseki endpoint, organized by protocol.

from jena_mcp.api_client import Api

api = Api(
    base_url="http://your-fuseki:3030",
    username="admin",
    password="admin",
    verify=False,
)

# SPARQL reads
results = api.query("ds", "SELECT * WHERE { ?s ?p ?o } LIMIT 10")

# Graph Store Protocol read (default graph as Turtle)
turtle = api.get_graph("ds")

# Server administration (read-only)
datasets = api.list_datasets()
info = api.server_info()
stats = api.stats("ds")

Build a client straight from the environment:

from jena_mcp.auth import get_client
api = get_client()        # reads JENA_* from the environment / .env

Writes¶

SPARQL updates and Graph Store writes mutate the dataset:

api.update("ds", "INSERT DATA { <urn:a> <urn:knows> <urn:b> }")
api.put_graph("ds", "<urn:a> <urn:p> <urn:o> .", graph="urn:g")   # replace
api.post_graph("ds", "<urn:a> <urn:p> <urn:o2> .", graph="urn:g") # merge

# Administration writes
api.create_dataset("ds2", db_type="tdb2")
api.backup("ds")
api.compact("ds", delete_old=True)

As a CLI¶

Run the MCP server directly with the jena-mcp console script (see Deployment for transports), or invoke the package as a module:

jena-mcp --help
python -m jena_mcp

The A2A agent server is the jena-agent console script:

jena-agent --mcp-url http://jena-mcp:8000/mcp --host 0.0.0.0 --port 8001

See Deployment → Agent server for the full agent configuration.