Local LLM Second Brain 🧠

Architecture: Telegram -> n8n -> Pi Agent -> Local Markdown

How it Works

1. Messages are sent via Telegram.
2. n8n catches the webhook, formats the payload, and sends it to Pi Agent via CLI.
3. Pi Agent reads the core system files (agent.md, rules.md, soul.md) and the user's message.
4. Pi Agent interacts with the local file system (PARA method) to read/write memories.
5. Every morning, a cron job/n8n schedule summarizes the previous day's session into /journals and clears the short-term memory.

Getting Started

To initialize the brain, send this exact message to the bot via Telegram: @agent execute bootstrap