863401402/she-love-me

What it solves

This project provides a tool to analyze chat history from WeChat and QQ to help users understand the dynamics of their romantic relationships. It quantifies relationship health using psychological frameworks (Attachment Theory, Gottman, and Sternberg's Triangular Theory of Love) and identifies "danger signals" such as gaslighting or love bombing.

How it works

The system operates as a "Universal Agent Skill" compatible with AI coding tools like Claude Code, Cursor, and GitHub Copilot. It follows a multi-step pipeline:

1. Data Extraction: It uses third-party tools (like weflow-cli, CipherTalk, or QQ Chat Exporter) to export chat logs into JSON or Markdown format.
2. Quantitative Analysis: A Python-based engine (stats_analyzer.py) calculates metrics such as response speed, message length ratios, and initiation frequency to create a stats.json file.
3. Layered Sampling: To avoid AI hallucinations with massive datasets, build_chat_history.py extracts specific windows of time (e.g., the last 30 days or high-conflict periods).
4. AI Interpretation: An LLM analyzes the combined statistics and sampled text to generate a psychological diagnosis, relationship status, and actionable advice.
5. Reporting: The results are output as a Markdown summary and a modern dark-themed HTML report.

Who it’s for

Individuals who want a data-driven, psychological analysis of their chat interactions with a specific contact to determine the relationship's status or identify toxic patterns.

Highlights

• Psychological Frameworks: Integrates Attachment Theory, Sternberg's Triangle, and Gottman's ratio for objective scoring.
• Danger Signal Detection: Uses a dual-threshold system (quantitative + textual) to flag risks like intermittent reinforcement or one-sided obsession.
• Local Processing: All data is processed locally; no chat logs are uploaded to external servers.
• Multi-Platform Support: Supports Windows WeChat and QQ, with import capabilities for macOS.
• Agent Integration: Can be triggered via simple commands (e.g., /she-love-me) within AI IDEs.
• Emoji Support: Includes tools to export and preview WeChat emojis locally.