📁 Subsystem Deep Dive

services/SessionMemory/

📁 services/SessionMemory/ 📄 7 files

Tutorial: SessionMemory

The Session Memory project implements an automatic, background note-taking system for AI conversations. It unobtrusively monitors the chat and, when specific activity thresholds are met, launches an isolated forked agent to summarize key information into a structured markdown file. This ensures the system maintains long-term context about tasks, files, and decisions without cluttering the main conversation flow or pausing the user's workflow.

flowchart TD A0["Post-Sampling Extraction Hook"] A1["Update Threshold Logic"] A2["Isolated Forked Agent"] A3["Prompt Construction & Templating"] A4["Shared State & Concurrency Control"] A5["Memory File Lifecycle"] A0 -->|"Checks update triggers"| A1 A0 -->|"Initializes and reads file"| A5 A5 -->|"Loads initial template"| A3 A0 -->|"Generates update instructions"| A3 A0 -->|"Dispatches extraction task"| A2 A0 -->|"Updates extraction status"| A4 A1 -->|"Reads config and history"| A4