writer-memory

To maintain character voice consistency across sessions, persistent project memory stores character tone, speech level, keywords, and taboo rules in a local memory.json file. This deterministic storage ensures dialogue validation logic can check new text against established voice profiles without resetting context.

Emotion-tagged scene planning records scenes with emotion tags and cut-level details to establish an emotional architecture. This approach prioritizes emotional consistency over plot mechanics, allowing writers to track character emotional progression and generate emotion-first synopses.

Yes, you can track relationship timelines by maintaining relationship evolution events between characters. Persistent memory records each relationship update as an event, building a chronological timeline that preserves continuity for character arcs across multiple writing sessions.

Generating an emotion-focused synopsis uses the stored scene emotion tags and character relationship data from persistent memory. It synthesizes the emotional trajectory of the narrative rather than summarizing plot points, producing a synopsis driven by character emotional arcs.

Fiction writing memory requires no external dependencies, relying solely on YAML-frontmatter-defined skill commands and deterministic storage to a local memory.json file. The entire memory architecture operates within the local project directory structure.

Dialogue validation checks new text against stored character voice attributes including tone, speech level, taboo words, and keyword consistency. The validation logic compares the input dialogue with the persistent profile in memory.json to flag inconsistencies.