Quickstart

Build gligabao, start the server, then remember and query your first facts.

1. Build and run

gligabao is pure Go and builds with CGO_ENABLED=0, so no C compiler is needed.

./run_build.sh                      # builds build/gligabao
export ANTHROPIC_API_KEY=...        # required for remember/query
./build/gligabao serve                # listens on :7878 by default

Prefer containers? docker compose up --build starts the server on localhost:7878 and the Python sidecar on localhost:8787, on one Docker network (the server talks only to the sidecar, which needs Modal tokens to proxy training calls to Modal). Set ANTHROPIC_API_KEY (and optionally OPENAI_API_KEY) in your shell first; data is kept in the gligabao-data volume.

Run ./run_tests.sh to run the test eval. Plain go build ./cmd/gligabao and go test ./... also work.

2. Remember something

curl -X POST localhost:7878/v1/gb/remember \
  -H 'content-type: application/json' \
  -d '{"user_id": "u1", "text": "Alice works at Acme and lives in Paris."}'

The response contains the artifact_id, the extracted entities, the edges with their temporal fields and any invalidated_edges. See Remember.

3. Query

curl -X POST localhost:7878/v1/gb/query \
  -H 'content-type: application/json' \
  -d '{"user_id": "u1", "query": "Where does Alice work?"}'

You get an answer plus the evidence (entities and edges) it relied on. See Query.

4. Inspect the graph

curl 'localhost:7878/v1/gb/datasets/main/graph?user_id=u1'

See Graph export.

Next steps

Configuration

API keys, database, vector store and prices.

Temporal facts

How contradicting facts are reconciled.