Install any skill in seconds. Free to start, no credit card required.
Get Started Free →当重要的故事元素(角色、情节线、伏笔)10章以上未出现时自动提醒 - 防止长篇小说中的'角色消失综合症'和遗漏的情节线
.claude/skills/microck-forgotten-elements-reminder/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 1% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -36% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -5% | 0% |
| case-11 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -26% | 0% |
防止长篇小说常见问题:
解决方案:后台监控,主动提醒已经很久未出现的元素。
监控:character-state.json中的所有角色
阈值:10章未出现
提醒示例:
⚠️ 角色提醒:
"配角李明"已经10章未出现(上次:第5章)
- 是否应该安排他再次出场?
- 还是这个角色的故事已结束?监控:plot-tracker.json中的活跃情节线
阈值:12章无进展
提醒示例:
⚠️ 情节线提醒:
"寻找父亲仇人"情节线已12章无进展
- 上次推进:第8章
- 当前状态:pending
- 建议:安排相关情节或标记为暂停监控:在前文埋下的重要伏笔
阈值:20章未回收
提醒示例:
⚠️ 伏笔提醒:
第3章提到"神秘盒子",至今未揭秘(已过23章)
- 读者可能已经忘记
- 建议:尽快回收或在对话中提及执行 `/write` 时,如果检测到遗忘元素:
📋 写作前检查...
⚠️ 发现3个被遗忘的元素:
1. 角色"张婶"已15章未出现
2. 情节线"寻宝"已13章无进展
3. 伏笔"神秘信件"已20章未揭秘
💡 本章是否考虑处理?执行 `/analyze` 时,生成完整报告:
## 遗忘元素报告
### 失踪角色(3个)
1. 张婶(15章未出场)
2. 李老板(11章未出场)
3. 王医生(10章未出场)
### 停滞情节线(2个)
1. 寻宝线(13章无进展)
2. 复仇线(12章无进展)
### 未回收伏笔(1个)
1. 神秘信件(20章未揭秘)
建议:
- 优先处理失踪角色(读者可能已忘记)
- 推进停滞情节线或标记为"暂停"
- 伏笔及时回收(否则成烂尾)"角色未出场提醒阈值改为15章"
"情节线停滞提醒阈值改为20章""张婶角色已完结,不再提醒"
"寻宝线有意暂停,30章后才推进,不提醒"/analyze 查看报告forgotten-elements-reminder = 你的记忆助手
✓ 自动监控角色/情节/伏笔 ✓ 超过阈值主动提醒 ✓ 防止长篇小说烂尾
长篇小说必备! 📝
本Skill版本: v1.0 最后更新: 2025-10-18
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 13,585 | 13,709 | +1% | 1 | 1 | 0% | 1,896 | 2,857 | +51% | 0 | 0 | — |
case-02 | fail→fail | 22,983 | 19,601 | -15% | 1 | 1 | 0% | 3,052 | 3,490 | +14% | 0 | 0 | — |
case-03 | fail→pass | 15,703 | 8,676 | -45% | 1 | 1 | 0% | 2,226 | 2,243 | +1% | 0 | 0 | — |
case-04 | fail→pass | 18,786 | 4,472 | -76% | 1 | 1 | 0% | 2,527 | 1,620 | -36% | 0 | 0 | — |
case-05 | pass→pass | 11,861 | 17,835 | +50% | 1 | 1 | 0% | 1,719 | 2,353 | +37% | 0 | 0 | — |
case-06 | pass→pass | 14,473 | 9,163 | -37% | 1 | 1 | 0% | 2,334 | 2,400 | +3% | 0 | 0 | — |
case-07 | pass→pass | 20,385 | 13,116 | -36% | 1 | 1 | 0% | 2,721 | 2,922 | +7% | 0 | 0 | — |
case-08 | pass→pass | 13,850 | 8,031 | -42% | 1 | 1 | 0% | 2,061 | 2,117 | +3% | 0 | 0 | — |
case-09 | pass→pass | 14,393 | 7,338 | -49% | 1 | 1 | 0% | 2,002 | 2,064 | +3% | 0 | 0 | — |
case-10 | fail→pass | 10,597 | 3,361 | -68% | 1 | 1 | 0% | 1,505 | 1,433 | -5% | 0 | 0 | — |
case-11 | fail→pass | 14,090 | 2,612 | -81% | 1 | 1 | 0% | 2,042 | 1,251 | -39% | 0 | 0 | — |
case-12 | fail→pass | 10,913 | 2,254 | -79% | 1 | 1 | 0% | 1,599 | 1,178 | -26% | 0 | 0 | — |
case-13 | fail→pass | 10,880 | 3,620 | -67% | 1 | 1 | 0% | 1,588 | 1,449 | -9% | 0 | 0 | — |
case-14 | fail→pass | 17,475 | 5,683 | -67% | 1 | 1 | 0% | 2,544 | 1,754 | -31% | 0 | 0 | — |
case-15 | pass→pass | 18,346 | 8,172 | -55% | 1 | 1 | 0% | 2,435 | 2,065 | -15% | 0 | 0 | — |
case-16 | fail→fail | 5,797 | 2,652 | -54% | 1 | 1 | 0% | 893 | 1,314 | +47% | 0 | 0 | — |
case-17 | fail→pass | 15,581 | 3,085 | -80% | 1 | 1 | 0% | 2,254 | 1,291 | -43% | 0 | 0 | — |
case-18 | pass→pass | 15,543 | 3,008 | -81% | 1 | 1 | 0% | 2,020 | 1,319 | -35% | 0 | 0 | — |
case-19 | fail→pass | 17,562 | 6,488 | -63% | 1 | 1 | 0% | 2,451 | 1,894 | -23% | 0 | 0 | — |
case-20 | fail→fail | 5,631 | 6,328 | +12% | 1 | 1 | 0% | 779 | 1,717 | +120% | 0 | 0 | — |
case-21 | pass→pass | 24,115 | 19,372 | -20% | 1 | 1 | 0% | 3,145 | 3,396 | +8% | 0 | 0 | — |
case-22 | pass→pass | 40,164 | 30,476 | -24% | 1 | 1 | 0% | 3,848 | 4,963 | +29% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +41 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.