Installation

riverbank supports multiple installation paths depending on your use case.

Python package (development)

uv (recommended)pip

uv venv && source .venv/bin/activate
uv pip install -e ".[dev]"

python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"

Extras groups

riverbank uses optional dependency groups to keep the core package lightweight:

Extra	Purpose	When to install
dev	pytest, ruff, mypy, testcontainers	Local development and testing
docs	mkdocs-material, mkdocstrings, mike	Building the documentation site
ingest	Docling, Instructor, spaCy, sentence-transformers	Real LLM extraction (production)
review	Label Studio SDK	Human review loop
orchestration	APScheduler, Prefect	Workflow orchestration
hardening	prometheus-client, aiobreaker, hvac	Production metrics, circuit breakers, Vault

Install multiple extras at once:

pip install -e ".[dev,ingest,hardening]"

Docker

The project ships a production Dockerfile:

docker build -t riverbank .

Or use the pre-built image:

docker pull ghcr.io/trickle-labs/riverbank:latest

Docker Compose (full stack)

The docker-compose.yml at the repo root brings up the complete stack for local development:

docker compose up -d

Services included:

Service	Purpose
postgres	PostgreSQL with pg-ripple, pg-trickle, pgvector
pg_tide	CDC relay sidecar
ollama	Local LLM runtime
langfuse	LLM observability dashboard
riverbank	Worker container

Kubernetes (Helm)

For production deployments, use the Helm chart:

helm install riverbank ./helm/riverbank \
  --set dbDsn="postgresql://..." \
  --set llmProvider=openai \
  --set llmApiKey="sk-..."

See the Helm chart reference for the full values.yaml documentation.

Verify your installation

After any installation path:

riverbank version
riverbank config

If you have a running PostgreSQL instance with the required extensions:

riverbank health