索引化记忆检索系统,支持按主题、日期、效用分类检索知识,避免重复学习,提高检索效率。
Scanned 9/11/2026
Install to Claude Code
npx -y skills add hiyenwong/ai_collection --skill indexed-memory --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Indexed Memory?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/hiyenwong-indexed-memory-ai-collection)More formats (shields.io, HTML) on the badges page.
---
name: indexed-memory
description: 索引化记忆检索系统,支持按主题、日期、效用分类检索知识,避免重复学习,提高检索效率。
model: haiku
tools:
- Read
- Write
- Glob
- Grep
color: "#3498DB"
---
# Indexed Memory - 索引化记忆检索
基于 Memex(RL) 论文的索引化记忆系统,支持高效检索和复用历史知识。
## 核心理念
> 人类记忆系统支持高效检索,AI 也应该如此。
索引化记忆解决以下问题:
- 知识库越来越大,难以快速定位
- 重复学习相同内容
- 重要知识被淹没在海量信息中
---
## 索引结构
```
knowledge/
├── index.md # 主索引
├── arxiv/ # arXiv 论文学习
│ └── index.json # 论文索引(arXiv ID → 文件路径)
├── skills/ # 技能库
│ └── index.json # 技能索引(技能名 → 文件路径)
└── daily/ # 日常记录
└── index.json # 日期索引
```
---
## 索引文件格式
### 论文索引 (arxiv/index.json)
```json
{
"2602.14697": {
"title": "E-SPL",
"file": "self-evolution-2026-03-06.md",
"keywords": ["self-evolution", "system-prompt", "RL"],
"utility": 0.9,
"lastAccessed": "2026-03-07"
},
"2505.22954": {
"title": "Darwin Gödel Machine",
"file": "self-evolution-2026-03-06.md",
"keywords": ["self-evolution", "open-ended"],
"utility": 0.85,
"lastAccessed": "2026-03-07"
}
}
```
### 技能索引 (skills/index.json)
```json
{
"ice-review": {
"file": "learned-skills.md",
"source": "arXiv:2401.13996",
"utility": 0.9,
"usageCount": 5,
"lastUsed": "2026-03-06"
},
"memory-retrieval": {
"file": "learned-skills.md",
"source": "arXiv:2601.03192",
"utility": 0.85,
"usageCount": 3,
"lastUsed": "2026-03-05"
}
}
```
---
## 检索流程
### 1. 论文检索
```bash
# 检查是否已学习某论文
grep "arXiv:XXXX.XXXXX" knowledge/arxiv/index.json
# 按关键词检索
grep -i "self-evolution" knowledge/arxiv/index.json
```
### 2. 技能检索
```bash
# 查找技能
grep "skill-name" knowledge/skills/index.json
# 按效用排序
jq 'to_entries | sort_by(.value.utility) | reverse' knowledge/skills/index.json
```
### 3. 日期检索
```bash
# 查找特定日期的学习内容
ls knowledge/arxiv/self-evolution-2026-03-*.md
```
---
## 效用评分机制
### 评分维度
| 维度 | 权重 | 说明 |
|------|------|------|
| 使用频率 | 30% | 被引用/使用的次数 |
| 成功率 | 40% | 应用成功的比例 |
| 时效性 | 30% | 最近使用时间 |
### 评分公式
```
utility = (usageScore * 0.3) + (successScore * 0.4) + (recencyScore * 0.3)
```
### 更新时机
1. **使用时** - 增加 usageCount
2. **成功应用时** - 更新 successRate
3. **定期审查** - 每周重新评估
---
## 去重机制
### 论文去重
```bash
# 学习新论文前检查
arxiv_id="2602.14697"
if grep -q "$arxiv_id" knowledge/arxiv/index.json; then
echo "已学习过此论文"
exit 0
fi
```
### 技能去重
```bash
# 创建新技能前检查
skill_name="new-skill"
if grep -q "\"$skill_name\"" knowledge/skills/index.json; then
echo "技能已存在,考虑更新而非创建"
exit 0
fi
```
---
## 维护操作
### 更新索引
```bash
# 添加新论文到索引
jq '. + {"new_arxiv_id": {"title": "...", "file": "..."}}' knowledge/arxiv/index.json > tmp.json && mv tmp.json knowledge/arxiv/index.json
# 更新效用评分
jq '.skill_name.utility = 0.95' knowledge/skills/index.json > tmp.json && mv tmp.json knowledge/skills/index.json
```
### 清理低效用内容
```bash
# 列出效用低于 0.5 的条目
jq 'to_entries | map(select(.value.utility < 0.5))' knowledge/skills/index.json
```
---
## 最佳实践
1. **学习前检索** - 避免重复学习
2. **定期更新效用** - 保持评分准确
3. **清理过时内容** - 保持知识库精简
4. **关联相关知识** - 建立知识图谱
---
## 输出格式
当用户需要检索知识时:
```markdown
## 检索结果
### 相关论文
| arXiv ID | 标题 | 效用 | 文件 |
|----------|------|------|------|
| 2602.14697 | E-SPL | 0.9 | self-evolution-2026-03-06.md |
### 相关技能
| 技能名 | 来源 | 效用 | 使用次数 |
|--------|------|------|---------|
| ice-review | arXiv:2401.13996 | 0.9 | 5 |
### 建议操作
- [ ] 应用 ice-review 技能
- [ ] 参考 E-SPL 论文优化系统提示
```
## Activation Keywords
- `indexed-memory`
- `indexed-memory`
- `indexed memory`
## Tools Used
- `exec`
- `read`
- `write`
- `edit`
## Instructions for Agents
1. Read the task description carefully
2. Follow the step-by-step process
3. Use the appropriate tools
4. Verify the results
## Examples
### Example 1: Basic Usage
**User:** <example user request>
**Agent:** <example agent response>
### Example 2: Advanced Usage
**User:** <example user request>
**Agent:** <example agent response>
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!