ns-canon

Original🇨🇳 Chinese
Translated
2 scriptsChecked / no sensitive code detected

Novel Canon Memory Maintenance Skill. Used to read and update established facts, plans, continuity, text indexes, style constraints, source boundaries, and visual indexes for a single NS project; maintains novel-studio/*.yaml and notes/*.md. Activated when users request to remember settings, save materials, update outlines, organize progress, handle end-of-chapter notes, correct continuity, or take over memory of an old novel. Initialization via ns-start.

11installs
Added on

NPX Install

npx skill4agent add honestman9527/novel-studio ns-canon

SKILL.md Content (Chinese)

View Translation Comparison →

NS Canon

Only maintains fact sources and progress; does not write new text or create unrequested new settings.
Read memory-schema.md for the complete structure.

Responsibilities

  • project.yaml
    : Project identity.
  • plan.yaml
    : Overall book scale, volumes, chapters, side stories, next steps.
  • memory.yaml
    : Characters, world, relationships, foreshadowing, genre modules.
  • continuity.yaml
    : Current status, events, unresolved plot threads, revision impacts.
  • index.yaml
    : Text entry paths, status, word count, sorting.
  • style.yaml
    : Writing style and chapter contracts.
  • research.yaml
    : Sources and fact boundaries.
  • publish.yaml
    : Root directory of text and publishing rules.
  • finish.yaml
    : Completion status and output index.
  • art.yaml
    : Visual consistency and file indexes.

YAML/Markdown Collaboration

  • The Markdown frontmatter of chapters and volumes serves as the connection layer between YAML memory and text files.
  • Frontmatter only includes identity, sorting, status, time, word count, and tags; long descriptions should be written in the Markdown text section or
    notes/*.md
    .
  • index.yaml.entries
    must be reconstructible from chapter frontmatter and audited word counts.
  • Volume
    _index.md
    maintains volume introductions, volume commitments, and chapter directories;
    plan.yaml
    maintains plans, and long texts should not be duplicated between them.
  • The optional
    ## End-of-Chapter Notes
    section in chapters is Markdown; if there are no end-of-chapter notes, update the canon directly from the text and context.
  • The total number of volumes, total chapters, number of side stories, and total word count target are only written in
    plan.yaml.scale
    .

Rules

  • New facts should only be written in one primary YAML field to avoid duplicate records.
  • Uncertain content should be written to
    continuity.yaml.loose_threads
    or
    research.yaml.open_questions
    .
  • After finishing a chapter, prioritize reading the optional
    ## End-of-Chapter Notes
    before updating
    index.yaml
    ,
    continuity.yaml
    , and
    memory.yaml
    .
  • Update
    finish.yaml
    only when a volume is completed, the entire book is finished, or deliverables change.
  • YAML is the source of facts; Markdown is for long-form explanations or display panels.
  • When adjusting the scale, first modify
    plan.yaml.scale
    and
    plan.yaml.volumes/extras
    , then synchronize the directory description in the volume
    _index.md
    .

Tools

  • Structure check:
    python skills/ns-canon/scripts/schema_doctor.py <novel-root>
    .
  • Continuity check:
    python skills/ns-canon/scripts/continuity_check.py <novel-root>
    .
  • Add
    --json
    when machine-readable results are needed; add
    --strict
    to continuity checks if unresolved plot threads should be treated as failure conditions.