What problem does it solve? Long-form AI fiction breaks down through context overflow, lost memory anchors, style drift, missing write-backs, and absent review cadence. This Skill enforces a disciplined production protocol where the agent writes prose while the Kinema engine handles registration, retrieval, and deterministic linting, keeping characters, foreshadowing threads, and world rules consistent across hundreds of chapters. ## Core Features & Use Cases - Batch writing protocol: Write in 10-chapter batches with a five-step per-chapter loop (brief, draft, three-gate self-check, save, write-back) and a mandatory seven-gate checkpoint review with a persisted batch report. - Structured story state: Register characters, props, scenes, volume outlines, foreshadowing threads, per-chapter digests, and end-of-chapter state snapshots via the python3 -m kinema novel CLI, with lint codes flagging gaps, drift, expired threads, and style deviation. - Takeover and continuation: Cold-start handoff of an existing book through a four-step intake (show, batch report, arcs, brief), plus an eight-step import checklist for user-supplied manuscripts. - Use Case: A user running a serialized web novel asks the agent to continue from chapter 66; the Skill pulls the chapter brief, writes chapters 66-75, runs the seven-gate review, files the batch report, and stops for user direction. ## Quick Start Ask the agent to continue writing your novel project by its project id, for example: continue my novel from the next chapter and run the checkpoint review when the batch of ten chapters is complete.
