Aperio

One brain. Every agent. Nothing forgotten.

A self-hosted memory layer and agent runtime for local or cloud AI.

Download Aperio-lite · Wiki · Discussions

What is Aperio?

Aperio keeps memories, conversations, and knowledge available to your AI agents.
It runs locally by default, stores data in SQLite, and exposes the same memory
through a web UI, terminal chat, and MCP-compatible tools.

Quick start

Easiest: Aperio-lite

For a guided installation, download Aperio-lite, unzip it, and double-click START.

From source

Requirements: Node.js 18+. Docker is optional;
SQLite is the default and needs no separate database server.

git clone --depth 1 -b dev https://github.com/BaiGanio/aperio.git
cd aperio
npm install
npm run start:local

Open http://localhost:31337 when the server is ready.

On first use, Aperio downloads the local AI engine and model. This may take a
few minutes and several gigabytes of disk space. You do not need to install
llama.cpp separately.

That is enough for a private local setup. The Web UI can configure most other
settings after startup.

Optional configuration

To configure a provider before starting, create .env from the example:

cp .env.example .env

Then set the values you need, for example:

AI_PROVIDER=llamacpp
LLAMACPP_MODEL=unsloth/gemma-4-E2B-it-qat-GGUF:UD-Q4_K_XL

Cloud providers require their own API key. Postgres, cloud embeddings, code/doc
graphs, encryption, path permissions, browser access, and other advanced options
are documented in the advanced guide and the
Aperio Wiki.

Terminal chat

Prefer a terminal? Start a second interface with:

npm run chat:local

It uses the same database and memories as the Web UI.

MCP and integrations

Aperio exposes memory, wiki, file, code-graph, document, and other tools over
MCP. See the MCP Tools Guide
for connection instructions and examples.

Help

Contributing

npm test

Read the testing guide and the
contributor/developer notes before changing the
runtime, database, configuration, or agent loop.

License

MIT