continual-learning

What problem does it solve?

This Skill addresses the issue of AI agents forgetting information between sessions, enabling them to retain and apply learned knowledge over time.

Core Features & Use Cases

• Persistent Memory: Implements a two-tier memory system (global and local) to store learnings.
• Automated Learning Capture: Hooks automatically capture tool usage patterns and failures to create new learnings.
• Agent-Native Storage: Allows agents to directly store learnings via SQL commands.
• Manual Memory Input: Supports human-readable, version-controlled memory files for explicit knowledge.
• Compaction Mechanism: Automatically prunes old or low-value learnings to manage memory size.
• Use Case: An AI agent repeatedly makes the same mistake when generating Terraform code. Continual learning captures this pattern, and the agent is prompted with the correct approach in future sessions, preventing recurrence.

Quick Start

Copy the continual-learning hook to your project's hooks directory and initialize the project memory database.