auto-memory-manager_skill

Overview

This skill builds an intelligent memory management system that gives an AI long-term memory. It automatically records every session, compiles nightly summaries, and extracts weekly highlights into a growing memory file. The system also supports real-time saving of critical events and automated cleanup to keep storage tidy.

How this skill works

The skill records structured session files on every conversation end, tagging topics, key information, todos, and decisions. A nightly job batches temporary sessions into a dated daily summary and removes temp files. A weekly task reads the last seven daily summaries to distill core insights into a persistent MEMORY.md while pruning stale daily files. Critical events can trigger immediate appends to long-term memory as a fail-safe.

When to use it

• Maintain a persistent history of user interactions for continuity across sessions
• Automate end-of-day summaries for fast review and handoff
• Consolidate weekly learnings into a single long-term knowledge file
• Capture urgent or high-value decisions immediately to avoid data loss
• Keep storage under control with scheduled cleanup of temporary files

Best practices

• Define a clear schema for session entries: topics, key_info, todos, decisions, emotion
• Schedule nightly and weekly jobs with reliable timezones and retries
• Enable real-time save for business-critical triggers (payments, releases, decisions)
• Set sensible retention (default 30 days) and review long-term memory periodically
• Use structured filenames and indexes to speed search and retrieval

Example use cases

• Personal AI assistant that remembers project history and decisions across weeks
• Customer support team that shares conversation records to onboard new agents faster
• Research groups compiling meeting notes into an evolving knowledge base
• Product teams tracking feature decisions and action items over time
• SaaS operators recording release and billing events to long-term memory