Dedup history
My personal collection of handcrafted AI agent skills — original prompting modules for code quality, knowledge exploration, and daily productivity.
npx -y skills add unix2dos/skills --skill dedup-historyAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 3 stars3 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Use when a content-generating skill needs deduplication against history records. Referenced as a sub-skill by skills that produce daily/random content.
SKILL.md
1.4 KB, as published. Nobody here has run it
通用去重流程
多个 Skill 共享的内容去重机制。
默认约定
| 参数 | 默认值 | 说明 |
|---|---|---|
| 历史文件 | {skill-name}_history.json | 与调用方 SKILL.md 同目录 |
| 上限条数 | 100 | 达到上限时移除最旧一条（FIFO） |
| 领域轮换排除数 | 3 | 排除最近 N 条的领域 |
调用方只需在自己的 SKILL.md 中声明 **REQUIRED SUB-SKILL:** Use dedup-history。如需覆盖默认值，显式声明即可。
流程
- 读取历史文件（与调用方 SKILL.md 同目录，若不存在则初始化为空数组
[]） - 检查候选内容是否已在历史记录中
- 若重复：重新选择（最多 3 次），3 次后输出提示"库存丰富但需要补充新内容"
- 生成内容
- 追加新记录到历史文件
- 如果记录数达到上限，移除最旧的一条（FIFO）
历史记录格式
[
{"title": "内容标题", "domain": "所属领域", "date": "YYYY-MM-DD"},
...
]
各 Skill 可扩展字段（如 author、region），但 title 和 date 为必须字段。
领域轮换
若 Skill 有领域池，从历史记录中获取最近 N 条的领域，排除后在剩余领域中随机选择，确保连续输出不重复领域。