@heixui/dsh-session-suite 0.4.0 → 0.4.2
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/README.md +17 -0
- package/README.zh.md +18 -0
- package/client.js +776 -113
- package/lib/api.js +15 -0
- package/lib/memory-engine.js +33 -4
- package/lib/memory-host.js +58 -4
- package/lib/service.js +154 -18
- package/package.json +1 -1
package/README.md
CHANGED
|
@@ -349,6 +349,23 @@ memory, or export paths.
|
|
|
349
349
|
two-layer agent-memory design, with measured numbers rather than claims — including
|
|
350
350
|
the parts that are deliberately not built and why.
|
|
351
351
|
|
|
352
|
+
## What changed in 0.4.1
|
|
353
|
+
|
|
354
|
+
A sidebar entry, plus three fixes for things that looked functional but were not.
|
|
355
|
+
|
|
356
|
+
| Area | Before | Now |
|
|
357
|
+
|---|---|---|
|
|
358
|
+
| Where to find it | Settings → Snapshots / Long-term memory, three steps away | A **sidebar entry** opens everything in place; the three duplicate Settings sections are removed |
|
|
359
|
+
| Memory count | Panel said 17, Settings said 51 — one question, two answers | The panel walks **every registered workspace**; each figure states what it counts |
|
|
360
|
+
| Archive restore | Clicked, nothing happened, the mark on disk never changed | Fixed an argument-shape mismatch between the two Host services; restore now takes effect |
|
|
361
|
+
| Long lists | The panel clipped its content; the wheel did nothing | The panel scrolls |
|
|
362
|
+
|
|
363
|
+
**Why the counts disagreed (the real 0.4.0 defect):** long-term memory has a *global*
|
|
364
|
+
store plus one store per workspace. `longterm.list` without a `cwd` returns only the
|
|
365
|
+
global store (17 rows); with a `cwd` it returns global plus that workspace. The panel
|
|
366
|
+
has no workspace of its own, so it showed the 17. It now enumerates every registered
|
|
367
|
+
workspace instead.
|
|
368
|
+
|
|
352
369
|
## What changed in 0.4.0
|
|
353
370
|
|
|
354
371
|
The memory engine was reorganised; the session features are unchanged.
|
package/README.zh.md
CHANGED
|
@@ -173,6 +173,24 @@ docs/ 需求与 API 笔记
|
|
|
173
173
|
**embedding 是可选增强,默认不下载任何东西。** 不装模型就是纯词法检索,功能完整;装了才有跨语言与改写召回。
|
|
174
174
|
实测效果与安装方式见 `docs/memory-architecture.md` 和 `tools/download-embedding.mjs`。
|
|
175
175
|
|
|
176
|
+
## 0.4.1 改了什么
|
|
177
|
+
|
|
178
|
+
新增侧边栏入口,修掉几个「看起来能用、实际没生效」的问题,并把界面上说不清的数字说清楚。
|
|
179
|
+
|
|
180
|
+
| 方面 | 改前 | 改后 |
|
|
181
|
+
|---|---|---|
|
|
182
|
+
| 入口位置 | 只能在 设置 → 会话快照 / 长期记忆 里找,三步才能到 | **左侧栏「会话套件」图标**,点开即用;设置页那三个重复入口已移除 |
|
|
183
|
+
| 记忆总数 | 面板显示 17,设置页显示 51 —— 同一句话两个数 | 面板遍历**所有已登记工作区**;每个数字都写明统计口径,并给出分工作区明细 |
|
|
184
|
+
| 归档恢复 | 点了没反应,磁盘上的标记纹丝不动 | 修了两个 Host 服务参数形状不一致的问题,恢复真正生效 |
|
|
185
|
+
| 快照数量 | 预算 10 张,列表却有 16 张,无从解释 | 区分「自有 10 张」与「继承 6 张」,继承项打徽章并说明为何不计入预算 |
|
|
186
|
+
| 隐式提取模型 | 只能选「关闭」和「用当前模型」,看不到自己配的模型 | 从 Host 的 llm 服务枚举真实可用模型;**默认关闭,不选就不消耗 token** |
|
|
187
|
+
| 长列表 | 配置区被五十条记忆挤到看不见 | 列表独立滚动,配置区始终可见 |
|
|
188
|
+
| 会话记忆 | 一个从未写入的残缺区块,旁边就是 dsh 自带的轨迹 | 已移除 —— 轨迹是权威且不可改,手写笔记只会误导 |
|
|
189
|
+
|
|
190
|
+
**关于记忆总数为什么会有多个数字**:长期记忆分「全局库」和「各工作区的库」。
|
|
191
|
+
`longterm.list` 不带 `cwd` 时只返回全局库,带上才是全局 + 该工作区。面板没有工作区概念,
|
|
192
|
+
于是只显示了全局那部分。现在面板会遍历所有已登记工作区,并在界面上直接列出每个库的条数。
|
|
193
|
+
|
|
176
194
|
## 测试
|
|
177
195
|
|
|
178
196
|
全部自测脚本都在临时目录里跑,不会碰真实的 DSH 数据。
|