Home / Manual
The My Thought Safe manual
Everything My Thought Safe does, explained the way you’ll meet it on screen. New here? Read the four Getting started pages in order and you’ll be writing in about ten minutes.
My Thought Safe is a private diary you can talk to. It runs an AI model on your own computer and keeps every word in one encrypted file that only you can open. It has no account, no cloud, and no network code. This manual covers version 0.10.0.
Two things to know before you start. First, there is no password reset. If you lose every way into a vault, nobody can open it, not even the author. Second, the AI model is a separate download: the app doesn’t include one. Both are explained step by step below.
Getting started#
InstallingWindows installer, portable zip and Linux AppImage.
Your first vaultPassphrases, security keys, and why there is no recovery.
Choosing a modelWhich GGUF file to get, and how the GPU and CPU share it.
Using the app#
Writing & conversationsThe chat screen, the context meter and per-conversation settings.
How memory worksSummaries, recall and the simulated infinite context.
Managing memoriesPin, search, edit, scope, export and import.
PersonasReusable characters and instructions for the AI.
Settings referenceEvery setting, tab by tab.
Keeping it safe#
Backups & integrityVerified backups, restoring, and checking for decay.
Privacy in practiceWhat stays hidden, what doesn’t, and good habits.
Help#
TroubleshootingFixes for the problems people actually hit.
Keyboard shortcutsThe handful of keys worth knowing.
How to read this manual#
- Words on buttons, tabs and checkboxes appear in bold inside quotes, like Back up now….
- Screenshots come from a demo vault on Windows. The Linux build looks the same.
- Where something has a limit, the manual says so. The complete list is on the security page.