VectorAdmin Documentation

The VectorAdmin monorepo is composed of the following core components:

• document-processor: A Flask application used to digest, parse, and embed documents.
• frontend: A ViteJS + React application for content management and UI.
• backend: A NodeJS + Express server that handles vector database interactions and management.
• workers: An InngestJS instance that manages long-running background tasks.
• docker: Orchestration files to run the entire architecture as a single instance.

When connecting to a Chroma instance running on the same machine as your Docker containers, use http://host.docker.internal:[CHROMA_PORT] as the URL.

Similar to Pinecone, a Sync Chroma Data button may appear on the organization homepage if existing data is detected in Chroma.

After starting the container, access the application at http://localhost:3001. You will be automatically redirected to an onboarding flow where you must:

1. Create your primary admin account.
2. Create an organization.
3. Establish a vector database connection.

To perform a complete initial setup of the project, including installing node modules for both the server and frontend, configuring environment files, generating the Prisma client, running migrations, and seeding the database, run the following command from the project root directory:

yarn setup

The fastest way to run VectorAdmin is to pull the latest pre-built image from Docker Hub. You must provide a valid DATABASE_CONNECTION_STRING for a Postgres database (RDS, remote, or local) before starting, as the vdbms database is created on boot using this string.

docker pull mintplexlabs/vectoradmin

docker run -d -p 3001:3001 \
--name vectoradmin \
-e SERVER_PORT="3001" \
-e JWT_SECRET="your-random-string-here" \
-e INNGEST_EVENT_KEY="background_workers" \
-e INNGEST_SIGNING_KEY="random-string-goes-here" \
-e INNGEST_LANDING_PAGE="true" \
-e DATABASE_CONNECTION_STRING="postgresql://vectoradmin:password@xxxxxxx:5432/vdbms" \
mintplexlabs/vectoradmin

To build the image from the local repository, follow these steps:

1. Clone the repository and navigate to the docker/ directory.
2. Copy the example environment file: cp .env.example .env.
3. Edit the .env file to update the following required variables:
   • JWT_SECRET: A random string for authentication.
   • DATABASE_CONNECTION_STRING: A valid Postgres connection string.
   • INNGEST_SIGNING_KEY: A random string.
4. Run the build command.

Note: If connecting to a Postgres instance running on your host machine from within the container, use host.docker.internal in the connection string.

git clone git@github.com:Mintplex-Labs/vector-admin.git
cd vector-admin
cd docker/
cp .env.example .env
# Edit .env with JWT_SECRET, DATABASE_CONNECTION_STRING, and INNGEST_SIGNING_KEY
docker-compose up -d --build vector-admin

To use Pinecone, navigate to your organization settings and provide your Pinecone configuration and keys.

After connecting, if you see a Sync Pinecone Data button on the organization homepage, it means there is existing data in your Pinecone index that can be pulled into VectorAdmin. The sync time depends on the number of embedded documents.

If you are transitioning from the old custom SQLite ORM setup to Prisma, you must migrate existing data. Navigate to the server directory and run the sqlite:migrate script.

Note: If you are a new user performing a fresh setup, this step is unnecessary as the standard setup already uses Prisma.