RecallVault
---
status: confirmed
created: 2026-10-27
updated: 2026-10-27
---
# RecallVault 文档索引
## 阅读顺序
按认知顺序排列，每一步都依赖前一步的结论：
1. **[竞品调研报告](research/competitive-analysis.md)** — GitHub 同类项目全景扫描，10+ 个竞品的能力矩阵和定位分析
2. **[痛点需求分析](requirements/pain-points-analysis.md)** — 4 个核心痛点的底层需求拆解 + 竞品覆盖度评估
3. **[隔离策略专题](architecture/isolation-strategy.md)** — 从"纯物理隔离"到"逻辑隔离+共享层"的决策过程
4. **[架构设计](architecture/design.md)** — 完整架构：目录结构、Frontmatter schema、MCP 工具签名、检索策略
5. **[自动存取路线图](architecture/roadmap.md)** — 基于 7 家竞品源码级深挖的实现方案：FTS5+RRF 融合检索、双时钟 frontmatter、onboard、Eyes 层、追溯索引（8 天实施）
6. **[工作方法](process/working-method.md)** — 开发哲学、MVP 分批、决策模型、验证策略
## 文档规范
### Frontmatter
每个文档顶部必须包含：
```yaml
---
status: draft | confirmed | archived
created: YYYY-MM-DD
updated: YYYY-MM-DD
---
```
- `draft`：待审阅，内容可能调整
- `confirmed`：已确认锁死，后续变更需显式记录
- `archived`：已废弃，保留供参考
### 引用格式
- 调研数据：`[cite:N]` 内联标注，文档末尾附 `## Sources` 节，每条格式为 `[cite:N] 标题 — URL`
- 架构决策：采用「决策 → 理由 → 备选方案 → 影响」四段式，记录在每个决策点
### 变更记录
- 状态从 `draft` → `confirmed`：更新 frontmatter `updated` 日期，在文档末尾追加一行确认记录
- `confirmed` 文档的内容变更：追加 `## 变更记录` 节，每条记录 `日期 | 变更内容 | 原因`
TDQS
Scored across 10 tools
Most tools have clearly distinct purposes, but update_memory explicitly sets archived and priority fields, overlapping archive_memory and set_memory_priority. remember versus update_memory also requires care, though the descriptions distinguish create/append from replacement.
The set mixes verb_noun names like list_memories, archive_memory, and delete_memory with bare verbs like remember and recall, plus noun-first names like health_check, memory_setup, and memory_import. It remains readable, but there is no single predictable convention.
Ten tools are well-scoped for a memory vault: CRUD/search, archive, priority, setup/import, and health check. Each tool has a plausible place in the lifecycle, with no obvious bloat.
The surface covers create/append, search/retrieve, list, update, delete, archive, priority, setup/import, and health. There is no dedicated get-by-id or bulk operation, but recall and list_memories cover retrieval and inspection well enough for core workflows.