What problem does it solve? Writers using AI for long-form fiction often struggle to make generated prose match the structure, pacing, and hook techniques of reference works they admire. This Skill analyzes a corpus of sample articles and distills a reusable imitation profile that guides subsequent novel generation without copying original expressions or settings. ## Core Features & Use Cases - Style Profile Synthesis: Analyzes .txt/.md/.markdown reference articles in a simulate/ folder using an architect-role LLM and writes a structured simulation_profile.json capturing structure, rhythm, and hook techniques. - Incremental Updates: Skips unchanged files via relative path plus SHA-256 fingerprinting, and merges new or modified articles into an existing profile without redundant LLM calls. - Profile Import: Imports a previously generated simulation_profile.v1 JSON, merging by corpus fingerprint and skipping duplicate sources. - Use Case: A web-novel author drops ten chapters of a favorite serialized novel into simulate/, runs the synthesis, and the resulting profile is injected into the Coordinator, Architect, Writer, and Editor agents so all future chapters borrow its pacing and hook patterns. ## Quick Start Place your reference articles into a simulate/ folder in the working directory and ask the agent to learn the writing style of these sample articles to synthesize an imitation profile.
