# GBrain AIO

GBrain AIO is a self-hosted ai app. GBrain AIO bundles GBrain — a self-hosted knowledge base and Model Context Protocol (MCP) server for AI agents — with PostgreSQL 17, pgvector, Caddy TLS, and a job supervisor in one container. It is open source and self-hosted, installable on Docker Hub and Unraid.

Canonical page: https://ourlocalhost.com/app/2448/gbrain-aio

## Description

GBrain AIO bundles GBrain — a self-hosted knowledge base and Model Context Protocol (MCP) server for AI agents — with PostgreSQL 17, pgvector, Caddy TLS, and a job supervisor in one container. Agents connect over HTTPS; Postgres is never exposed. Ideal for agent frameworks like Hermes Agent, OpenClaw, and Codex that need persistent, queryable memory without managing separate services.

## Platform availability

Available on 1 of 10 platforms: Unraid.

## Health score

54/100 (Fair), based on 3 of 4 factors.
- Maintained: 91 (last commit 11 days ago)
- Popular: 10 (2 GitHub stars)
- Easy to install: 35 (1 install platform, install notes, Docker image)

## Links

- Docker Hub: https://hub.docker.com/r/dub19/gbrain-aio
- GitHub: https://github.com/wgross19/gbrain-aio
- Unraid: https://ca.unraid.net/apps/gbrain-aio-0jujfb70o0ck7b?q=gbrain-aio

## Common questions

### Do I need to configure a reverse proxy for gbrain-aio?

No; the container includes Caddy TLS on port 3132 in front of GBrain's loopback service on 3131, and Postgres is never exposed.

### How do I avoid being locked out of admin on first boot?

Set GBRAIN_ADMIN_BOOTSTRAP_TOKEN or save the admin bootstrap token printed during first run, because missing it can lock you out of the admin login.

### Which folder should I use for the GBrain root?

Use a durable folder for the GBrain root, not a temporary runtime folder, because it holds your saved memory and may disappear if it does not survive restarts.

### Does gbrain-aio include the embedding model, or do I need another service?

You need a local Ollama instance, and only ollama:embeddinggemma at 768 dimensions is supported; point OLLAMA_BASE_URL at it, and a GPU is strongly recommended for large brains.

### How should I verify the brain works before adding real data?

Save a harmless test fact and recall it from a fresh conversation. If it fails, check the brain root folder, file permissions, and setup steps before adding real data.

---
Source: Our Local Host (https://ourlocalhost.com/app/2448/gbrain-aio), the self-hosted and open-source app directory.
