What problem does it solve? It turns scattered personal data—WeChat/QQ chat exports, diaries, social media posts, and photos—into a structured, invocable AI persona that thinks and speaks like you, without manual prompt engineering. ## Core Features & Use Cases - Multi-source ingestion: Parses WeChat (WeChatMsg, LiuHen, PyWxDump), QQ (txt/mht), social media text, and photo EXIF metadata to extract speech patterns, values, and life timelines. - Dual-layer generation: Produces Part A (Self Memory: experiences, values, habits, relationships) and Part B (Persona: 5-layer personality model from hard rules to interpersonal behavior), combined into a runnable SKILL.md. - Evolution and versioning: Supports appending new materials, conversational corrections ("I wouldn't say that"), automatic version backups, and rollback. - Use Case: Export your WeChat history, run the intake flow, and get a /{slug} command that replies in your tone, catchphrases, and decision logic. ## Quick Start Ask the agent to create a self skill from your chat history by saying: help me distill myself into a skill using my exported WeChat chat logs.