Extracts AES-256 encryption keys from WeChat process memory on Windows and macOS to decrypt SQLCipher-encrypted chat databases. Demonstrates platform-specific memory scanning (pymem on Windows, lldb on macOS) and PBKDF2 key derivation. The pattern of live memory extraction from a running desktop application is reusable for any project needing to access encrypted local data from WeChat or similar apps, though the specific offsets and SQLCipher parameters require version-specific maintenance.

Parses decrypted WeChat SQLite databases and macOS iMessage chat.db to extract, classify, and format chat messages. Supports multiple export formats (text, JSON, CSV) and classifies messages into long, conflict, sweet, and daily categories using keyword matching. The parser's ability to handle different database schemas across platforms and versions, along with its conversation thread extraction, makes it a reusable asset for chat data analysis and digital persona construction projects.

Serves as the entry point and orchestration script for the entire ex-skill system. Defines a five-step workflow (intake, data import, analysis, preview, file writing) and supports incremental updates, conversational corrections, and version rollback. While tightly coupled to this project's prompt pipeline, the structural pattern of a meta-skill that generates other skills following the AgentSkills standard is a novel, reusable architecture pattern for building generative persona systems.