kropdx/reflection-engine

What it solves

It provides a way to uncover personal blind spots, contradictions, and behavioral patterns that are difficult to see from the inside by leveraging the vast amount of personal data an AI assistant already possesses.

How it works

It is a single, complex prompt delivered as a markdown file. When attached to an AI assistant (like ChatGPT, Claude, or Gemini) with memory and chat history enabled, the model analyzes the user's entire history of conversations, uploaded files, and stored memories to generate a candid, evidence-grounded portrait of the user. The process produces 22 direct answers, each backed by evidence, a confidence score, and a concrete action item.

Who it’s for

Individuals who use AI assistants for a wide range of personal topics (relationships, health, money, etc.) and want a sharp, objective analysis of their own patterns and trajectories.

Highlights

• Privacy-first design: As a prompt rather than a software service, data never leaves the existing AI provider's environment.
• Evidence-based: Every insight is guided by evidence from the user's chat history and memory.
• Reasoning-heavy: Designed for high-reasoning models to avoid generic outputs.
• Pattern recognition: Focuses on contradictions and "expensive habits" rather than simple summaries.