self-correction-loop

What problem does it solve?

This Skill addresses the issue of AI models repeatedly making the same errors by establishing a robust system for capturing, generalizing, and storing user corrections as reusable knowledge.

Core Features & Use Cases

• Correction Capture: Automatically detects and acknowledges user corrections.
• Generalization: Transforms specific feedback into broad, class-level rules.
• Memory Management: Stores generalized rules in MEMORY.md, categorized for retrieval and deduplicated.
• Proactive Application: Reviews stored memory at the start of each session to prevent recurring mistakes.
• Use Case: If a user corrects the AI for using DateTime.Now instead of TimeProvider in a specific instance, the AI will generalize this to a project-wide rule: "Always use TimeProvider instead of DateTime.Now/UtcNow across all modules."

Quick Start

When a user corrects the AI's output, acknowledge the correction, generalize the lesson, and store it as a reusable rule in MEMORY.md.