How It Works

• Set a Username: Enter your unique handle in the plugin configuration. On first use the nickname will be saved in the name database.
• Talk to the Model: Ask the AI to "Check the post db for new messages" or "Send a private note to tupik".
• Automated Flow: The model parses parameters, authenticates anonymously, and securely syncs JSON text streams with the cloud.
• Database location: United States (us-central1) segment of Google servers.

Installation

or Hub: tupik/lm-post click run

Ideas for future development

Post me these in there to tupik or admin.
An idea for how to use this database: integrate message checking into the preprocessor so that they are included in the LLM context. This way, a set of custom settings can be stored. Why online and not locally? To decouple from the device. And for syncing between them.

Firebase RTDB -- Rules of database

// 1. Username Registration Node
"usernames": {
"$username": {
".read": "auth != null",
// You can create a username if it doesn't exist. You can only overwrite it if you own it.
".write": "auth != null && (!data.exists() || data.val() === auth.uid)"
}
},

// 2. the Chat data base
"lm-chat": {
".indexOn": ["ts"],
// Only authorized users can read the chat logic (our plugin makes them anonymous users)
".read": "auth != null",
"$message_id": {

// A hacker can't delete the entire lm-chat; they can only CREATE a new entry ($message_id)
// and only if the 'uid' field inside the entry matches their anonymous ID
".write": "auth != null && (!data.exists() || data.child('uid').val() === auth.uid)",
".validate": "newData.hasChildren(['from', 'to', 'text', 'ts', 'uid'])"
}
}