writer-memory

Persistent writer memory tracks character voice consistency by storing arcs, speech levels, keywords, and taboo words in a local JSON file. This maintains dialogue tone and baseline emotions across sessions for continuous reference.

The best way to track relationship timelines is by recording relationship types, statuses, and power dynamics in a dedicated memory system. This allows you to map an evolution timeline and reference how character bonds change throughout the story.

You validate dialogue against established character memory by cross-referencing the text with stored speech levels, keywords, and emotional triggers. This validation process checks if new dialogue aligns with the character's recorded voice profile.

Yes, scene memory tracks scene structure by storing cuts, narration tone, emotional tags, and before/after character changes. This scene memory supports emotional engineering by mapping exactly how each scene impacts character development.

Korean natural-language query support works by searching the stored writer memory context using conversational Korean prompts. This allows writers to retrieve character details, relationship timelines, and theme tracking data without using strict command syntax.

To initialize a local JSON file for fiction context tracking, you need to run an initialization command that creates a project directory and a memory file. This sets up the deterministic storage structure for loading and updating your writer state.