novel-outline-reviewer-d90347 — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited novel-outline-reviewer-d90347 (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Systematically evaluate novel content. When reviewing chapters 1–3, always run the First-Chapter Editorial Gate first — if it returns NOT READY, halt and prompt the user before proceeding. For all other chapters, go straight to the Nine-Dimension Review (D1–D9).
Load context in parallel:
characters/*.md (personality, speech patterns, emotional baseline, growth arc)plot/outline.md (chapter hooks, timeline, plot beats)worldbuilding/world.md (setting rules, cultural context)仅当审查内容为第1–3章时运行。 直接对正文进行审查,无需预加载角色卡或大纲。
⚠️ HALT RULE(强制执行) 若本门控总评为 NOT READY,立即停止,输出门控报告后向用户提示: _"前三章过稿审核未通过,建议先修改以下问题,再进行深度九维审查。是否现在继续后续审查?"_ 等待用户明确确认后,再决定是否进入 D1–D9。
检查第一章前三行 / 首段:
判定标准:
编辑筛稿逻辑:80% 读者前三行看不到冲突直接划走,慢热开篇等于主动出局。
逐 500 字扫描,每区间至少存在一个钩子类型:
| 钩子类型 | 示例 |
|---|---|
| 神秘线索 | 出现无法解释的细节或异常现象 |
| 角色反常行为 | 人物做出与处境明显不符的选择或反应 |
| 未知危机 | 读者感知到威胁,但主角尚未察觉 |
| 未解谜团 | 提出问题,暂不给答案 |
判定: 任意 500 字区间无钩子 = NEEDS ATTENTION;连续两个区间无钩子 = CRITICAL。
绘制全章情绪走势:
常见失误:
主角是否具备清晰的人设反差(反差感)?
判定: 主角无辨识度反差、形象扁平 = NEEDS ATTENTION;与同赛道主角高度同质 = CRITICAL。
扫描以下水文信号:
判定: 单章内出现 2 处以上水文 = NEEDS ATTENTION;节奏整体拖沓 = CRITICAL。
验证故事基础公式:普通人 + 异常事件 + 紧迫生死困境
判定: 公式缺少任意一项 = NEEDS ATTENTION;三项均缺 = CRITICAL。
逐章检查末段:
| 章节 | 结尾钩子类型 | 强度 |
|---|---|---|
| 第1章 | [突发危机 / 关键反转 / 悬而未决的问题 / 其他] | STRONG / WEAK / FAIL |
| 第2章 | [同上] | STRONG / WEAK / FAIL |
| 第3章 | [同上] | STRONG / WEAK / FAIL |
判定: 任意章节在平静解决中结束,无前向拉力 = FAIL,须加入追更钩子。
每项必须给出 PASS / FAIL,不得笼统评价。
| 检查项 | 规则 |
|---|---|
| 真实地名 | 禁止出现可识别的真实地名(城市、街道、学校、机构等) |
| 暴力描写 | 死亡/伤害场景须用意外替代(车祸、失足、突发疾病),禁止血腥分尸、投毒细节 |
| 未成年限制 | 未成年角色禁止恋爱描写和肢体接触 |
| 视角节奏 | 每300字内有视角焦点切换或场景动作,避免长段单视角内心独白 |
| 对话伏笔 | 重要对话是否暗藏可供后期呼应的细节,而非纯功能性推进 |
任一项 FAIL = 投稿前必须修改。
门控总评:
仅在门控通过(READY 或用户确认继续)后运行。 Work through ALL nine dimensions. Never skip one because "it seems fine."
Ask for EACH scene:
Common failures: Character makes uncharacteristically impulsive choice; timeline gaps that don't add up; object/detail appears before it's introduced.
Compare behavior in chapter against character file:
| Check | Question |
|---|---|
| Speech pattern | Does dialogue match their established vocabulary and verbal tics? |
| Emotional response | Would this person react this way given their background? |
| Growth arc | Is any change in behavior earned by prior events, not sudden? |
| Relationship logic | Do interactions reflect established relationship dynamics? |
Red flag: A reserved character suddenly opens up with no transitional scene; a logical character acts on pure impulse without internal justification.
Compare every scene detail against worldbuilding/world.md:
| 检查项 | 检查问题 |
|---|---|
| 地理与空间 | 场景位置、距离、建筑布局与设定吻合?(如场所各楼层功能区、标志性空间等细节) |
| 时代背景 | 技术水平、社会现象、流行事物与故事设定年代相符? |
| 经济与生活 | 人物消费能力、收入水平与其背景设定匹配?次要角色的生活细节写实? |
| 规则与惯例 | 故事核心场所的特有规则(如准入门槛、特殊制度等)在各章执行是否前后一致? |
| 物候与环境 | 季节描写(植物、天气、物候)与故事时间线对应? |
| 群像生态 | 固定配角(邻居、伙伴、相关人物等)的出场与其设定角色相符? |
常见失误:
Maintain a running foreshadowing ledger:
PLANTED: [detail/item/line] in Chapter X
PAID OFF: [resolution] in Chapter Y — or — STATUS: OPENCheck:
Show-don't-tell audit: For every direct emotion statement ("she felt sad"), find a behavioral or sensory replacement.
| Weak | Strong |
|---|---|
| "She felt lonely" | "She reheated the same tea three times without drinking it" |
| "He was nervous" | "He checked the door twice, then checked it again" |
Five-sense check — does the scene use more than just visual description?
Emotional beat map:
Dialogue naturalness: read lines aloud — would a real person say this?
This dimension is MANDATORY and must produce a PASS/FAIL verdict, not just "NEEDS ATTENTION."
Check each item below. Any single violation = CRITICAL, must be flagged and fixed before publication.
| 红线类别 | 检查内容 |
|---|---|
| 政治与地域 | 无政治敏感内容、无地域歧视表达、无民族/宗教违规描写,不涉及敏感历史事件或敏感人物 |
| 暴力与违法 | 无暴力、血腥、惊悚、恐怖情节,无打架斗殴场景,无违法犯罪或灰色产业描写 |
| 色情与两性 | 无低俗色情、无露骨性描写,情感线方向纯爱治愈、积极健康,不含不良两性关系导向 |
| 消极心理导向 | 无消极厌世情绪、无自杀/自残内容、无躺平摆烂导向,不渲染焦虑,不传播负面情绪;若涉及抑郁/心理健康话题,须以康复和积极连接为方向 |
| 信息真实性 | 无虚假信息、无不良引导;乡村生活、职场情节均为现实正向描写,符合公序良俗 |
| 版权原创性 | 所有设定、情节、人物均为原创,未抄袭、未改编违规素材,对话及场景不与已知作品高度相似 |
注意: 抑郁症等心理健康话题本身不违规,但描写方向必须指向康复与连接,不得渲染绝望或美化轻生。
This dimension evaluates whether the chapter delivers memorable moments that make readers want to keep reading, share quotes, or reread scenes.
Check each category:
| 类型 | 检查问题 |
|---|---|
| 名场面 | 是否有让人"刷到这段必停下来"的画面或对话?场景是否有独特的画面感和记忆点? |
| 高光时刻 | 角色有没有说出或做出让读者拍案叫绝、想截图的瞬间? |
| 爽点 | 有没有让读者"终于来了"的情节释放?积压的张力是否有出口? |
| 撒糖时刻 | 情感线有没有甜蜜细节?男女主的互动有没有让读者"磕到了"的moment? |
| 群像魅力 | 配角是否有存在感?多人场景中每个角色是否有自己的反应和声音? |
| 温馨时刻 | 有没有让读者心里一暖、想到"原来生活可以这样"的画面? |
密度检查: 每3000字至少应有1个以上高光类型的时刻。若整章无任何一类,即使逻辑通顺,读者也会感到"平"。
常见问题:
For each dimension, report:
## [Dimension Name]
STATUS: PASS / NEEDS ATTENTION / CRITICAL
Issues found:
- [Chapter X, Scene Y]: [specific problem] → Suggested fix: [concrete rewrite direction]
Strengths worth preserving:
- [What's working well]D8 special rule: Must list each of the 6 red-line categories with PASS or FAIL. Any FAIL must include the exact offending text and a concrete rewrite direction. Do not summarize as "looks fine" without item-by-item confirmation.
Gate special rule (first 3 chapters only): Must list each of the 7 sub-checks with PASS / WEAK / FAIL. End with a clear READY / NEEDS REVISION / NOT READY verdict. Any FAIL in Submission Compliance automatically blocks the READY verdict. If NOT READY, halt here and prompt the user before proceeding to D1–D9.
End with a Priority Fix List (max 5 items, ranked by impact on reader experience). Priority order: Gate NOT READY items → D8 CRITICAL → D9 CRITICAL → other dimensions.
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.