#Quickstart

Get a Legra node running and query it in under five minutes.

#1. Install a node

A node is a local Legra instance with its own identity and default workspace.

sudo legra install --apt --node-only

This installs Legra daemons on the default port block, generates cryptographic identities, and creates a default workspace. The gateway HTTP port is 4242 on every platform.

Each machine runs at most one host; to run multiple hosts, use multiple machines (physical or virtual).

#2. Start the server

legra node serve

By default, the gateway listens on all network interfaces at 0.0.0.0:4242. This means other people and other nodes on the network can connect to it — which is safe and recommended, since all workspace data is end-to-end encrypted.

If you prefer to restrict access to your own machine only, bind to localhost:

legra node serve --host 127.0.0.1

The server accepts both HTTP requests and connections from other Legra nodes on the same port.

For quick experiments, you can run in ephemeral mode:

legra node serve --ephemeral

Ephemeral mode keeps all data in memory instead of writing it to disk. Data is lost when the server stops. Because everything must fit in RAM, this mode is best suited for small datasets and short-lived experiments.

#3. Load sample data

Create a file called people.ttl:

@prefix : <http://example.org/> .

:Alice  :name "Alice" ;
        :knows :Bob .

:Bob    :name "Bob" ;
        :knows :Alice .

Load it into the server:

legra graph upload people.ttl

The upload command also accepts a directory. It recursively scans for RDF files (Turtle, N-Triples, RDF/XML) in all subdirectories and loads them in one go:

legra graph upload ./data/

If you omit the path, it scans the current directory:

legra graph upload

See the Loading Data guide for more options.

#4. Run a query

The easiest way to query is with the legra CLI, which uses the native protocol for the best performance. The legra graph query command accepts a SPARQL query string.

legra graph query "SELECT ?name WHERE { ?person <http://example.org/name> ?name }"

Output (JSON by default):

{
  "results": {
    "bindings": [
      { "name": { "type": "literal", "value": "Alice" } },
      { "name": { "type": "literal", "value": "Bob" } }
    ]
  }
}

For tab-separated output:

legra graph query "SELECT ?name WHERE { ?person <http://example.org/name> ?name }" --format tsv

You can also query via HTTP using any tool that speaks the SPARQL Protocol, such as curl:

curl -X POST http://127.0.0.1:4242/api/v1/sparql/query \
  -H "x-legra-api-key: $EKG_SPARQL_AUTH_KEY" \
  -H "Content-Type: application/sparql-query" \
  -d "SELECT ?name WHERE { ?person <http://example.org/name> ?name }"

If your node is configured for JWT-based auth instead of API token auth, send Authorization: Bearer <jwt> instead of x-legra-api-key.

See the HTTP API reference for details.

#5. Create a branch

Branches let you make changes without affecting the main timeline:

legra graph branch create experiment

Query against the new branch:

legra graph query "SELECT * WHERE { ?s ?p ?o }" --branch experiment

Load data into it:

legra graph upload more-data.ttl --target-branch experiment

List all branches:

legra graph branch list

#Next steps

• Loading Data — formats, named graphs, directory loads
• Querying — SPARQL features, output formats, query plans
• Branching — merge workflows, delta queries
• Workspaces — encryption, membership, key management
• CLI Reference — full command reference