Home
Welcome! This wiki is the complete guide to using Apricity — what every panel does, how each feature works, and how to configure and secure it. Everything here is based on how the current code actually behaves.
Apricity is the window onto your MemPalace. MemPalace gives an AI assistant a structured, durable memory — organised into wings → rooms → drawers (memories) plus a knowledge graph of facts — all stored locally. Once you have hundreds of memories, you need a way to see, search, and curate them without dropping into SQL or living inside an MCP shell. That's Apricity.
It is not a memory system of its own. It reads and writes an existing MemPalace install on the same machine, through the official mempalace Python package. If MemPalace isn't installed, Apricity has nothing to show.
Vocabulary: a wing is a top-level area (often a project), a room is an aspect within a wing (e.g.
decisions,backend), and a drawer is a single memory. Apricity's UI calls drawers "memories."
| I want to… | Go to |
|---|---|
| Find a memory fast | Browsing & Searching |
| Write or edit a memory | Writing & Editing |
| Delete something (and undo it) | Deleting & Recovering |
| Work with facts & links | Knowledge Graph & Tunnels |
| Know when memories change | Notifications |
- 🏠 Local-first. Binds to
127.0.0.1. No telemetry, no cloud, no accounts beyond your own local login. - 🪶 Zero runtime dependencies. A single standard-library Python server + plain HTML/CSS/JS. No
npm, no Docker, no build step. - 🛡️ Safe by construction. Every write goes through the official
mempalacepackage; every delete and edit is snapshotted and recoverable; ETag concurrency control prevents lost edits. - ⚡ Fast. Reads the SQLite/Chroma backends directly; a localStorage cache, gzip, and serve-time minification make reloads feel instant.
- Installation & First Run — get it running and create your login.
- Tour of the Interface — learn the three-pane layout in two minutes.
- Keyboard Shortcuts — the handful of keys that make it fast.