Skip to content

Hypergraph is more powerful. Transform unstructured text into structured knowledge with LLMs. Graphs, hypergraphs, and spatio-temporal extractions — with one command.

docs: show pickle-free index.json only for model/list

Frankie-Xu194b5afe47fb
diff --git a/docs/en/cli/commands/build-index.md b/docs/en/cli/commands/build-index.md
index 39953e1..b24bcbd 100644
--- a/docs/en/cli/commands/build-index.md
+++ b/docs/en/cli/commands/build-index.md
@@ -105,18 +105,30 @@ he build-index ./ka/ -f
## Index Storage
-The index is stored in the knowledge abstract directory:
+The index is stored in the knowledge abstract directory. Layout depends on the Auto-Type:
+
+**AutoModel / AutoList** (pickle-free JSON since v0.10.1):
+
+```
+./ka/
+├── data.json
+├── metadata.json
+└── index/
+ └── index.json # vectors + documents, no pickle
+```
+
+**Graph family / AutoSet / AutoDocument** (still OMem — not the v0.10.1 JSON path):
```
./ka/
├── data.json
├── metadata.json
-└── index/ # Index directory
+└── index/
├── index.faiss # FAISS vector index
└── docstore.json # Document store mapping
```
-Load an existing `index/` only from Knowledge Abstract directories you created or fully trust. Since v0.10.1 indexes are JSON (safe to load); legacy pickle indexes (<= v0.10.0) are deserialized on load — rebuild with `--force` to migrate.
+Load an existing `index/` only from Knowledge Abstract directories you created or fully trust. For AutoModel / AutoList, v0.10.1+ writes `index.json` and loads it without executing code; leftover `index.faiss` / `index.pkl` from <= v0.10.0 still deserialize on load — rebuild with `--force` to migrate. Graph / set / document indexes stay on OMem, so the trust warning still applies.
---
diff --git a/docs/en/cli/commands/parse.md b/docs/en/cli/commands/parse.md
index e0edf9c..0f3db37 100644
--- a/docs/en/cli/commands/parse.md
+++ b/docs/en/cli/commands/parse.md
@@ -151,10 +151,14 @@ cat document.md | he parse - -t general/biography_graph -o ./output/ -l en
│ ├── created_at # Creation timestamp
│ └── updated_at # Last update timestamp
└── index/ # Vector search index (if built)
- ├── index.faiss
- └── docstore.json
```
+Index files depend on the Auto-Type:
+
+**AutoModel / AutoList** (pickle-free JSON since v0.10.1): `index/index.json`.
+
+**Graph family / AutoSet / AutoDocument** (still OMem): `index/index.faiss` + `index/docstore.json`.
+
---
## Language Support
diff --git a/docs/en/python/guides/saving-loading.md b/docs/en/python/guides/saving-loading.md
index 19f5923..f6d8de6 100644
--- a/docs/en/python/guides/saving-loading.md
+++ b/docs/en/python/guides/saving-loading.md
@@ -30,6 +30,18 @@ result.dump("./my_ka/")
### Save Structure
+**AutoModel / AutoList** (pickle-free JSON since v0.10.1):
+
+```
+./my_ka/
+├── data.json # Extracted knowledge
+├── metadata.json # Extraction info
+└── index/ # Search index (if built)
+ └── index.json # vectors + documents, no pickle
+```
+
+**Graph family / AutoSet / AutoDocument** (still OMem — not the v0.10.1 JSON path):
+
```
./my_ka/
├── data.json # Extracted knowledge
@@ -54,7 +66,7 @@ result.dump("./my_ka/")
## Loading Knowledge Abstracts
!!! warning "Trusted directories only"
- Since v0.10.1, indexes are stored as JSON (`index.json`) and load without code execution. Legacy KAs (<= v0.10.0) keep pickle-based indexes, which ARE deserialized on load — only call `load()` (or CLI/MCP commands that load a KA) on directories you generated yourself or fully trust. Rebuild with `he build-index --force` to migrate a legacy KA.
+ Only call `load()` (or CLI/MCP commands that load a KA) on directories you generated yourself or fully trust. Since v0.10.1, **AutoModel / AutoList** indexes are stored as JSON (`index.json`) and load without code execution. Legacy model/list KAs (<= v0.10.0) keep pickle-based indexes (`index.faiss` / `index.pkl`), which ARE deserialized on load — rebuild with `he build-index --force` to migrate. Graph / set / document KAs still persist through OMem and are not on the v0.10.1 JSON path, so the same trust warning still applies.
### Basic Load
diff --git a/docs/zh/cli/commands/build-index.md b/docs/zh/cli/commands/build-index.md
index 53373b7..8bf90e3 100644
--- a/docs/zh/cli/commands/build-index.md
+++ b/docs/zh/cli/commands/build-index.md
@@ -105,18 +105,30 @@ he build-index ./ka/ -f
## 索引存储
-索引存储在知识库目录中:
+索引存储在知识库目录中。布局取决于 Auto-Type:
+
+**AutoModel / AutoList**(自 v0.10.1 起为无 pickle 的 JSON):
+
+```
+./ka/
+├── data.json
+├── metadata.json
+└── index/
+ └── index.json # 向量 + 文档,无 pickle
+```
+
+**图谱家族 / AutoSet / AutoDocument**(仍由 OMem 落盘,不是 v0.10.1 的 JSON 路径):
```
./ka/
├── data.json
├── metadata.json
-└── index/ # 索引目录
+└── index/
├── index.faiss # FAISS 向量索引
└── docstore.json # 文档存储映射
```
-请只加载自己生成或完全信任的知识库目录中的 `index/`。自 v0.10.1 起索引为 JSON(加载安全);旧版 pickle 索引(<= v0.10.0)加载时会反序列化——用 `--force` 重建即可迁移。
+请只加载自己生成或完全信任的知识库目录中的 `index/`。对 AutoModel / AutoList,v0.10.1+ 写入 `index.json` 并在加载时不执行代码;<= v0.10.0 残留的 `index.faiss` / `index.pkl` 加载时仍会反序列化——用 `--force` 重建即可迁移。图谱 / set / document 索引仍走 OMem,因此这条信任警告仍然成立。
---
diff --git a/docs/zh/cli/commands/parse.md b/docs/zh/cli/commands/parse.md
index fa1d852..671ba54 100644
--- a/docs/zh/cli/commands/parse.md
+++ b/docs/zh/cli/commands/parse.md
@@ -146,10 +146,14 @@ cat document.md | he parse - -t general/biography_graph -o ./output/ -l zh
│ ├── created_at # 创建时间戳
│ └── updated_at # 最后更新时间戳
└── index/ # 向量搜索索引(如已构建)
- ├── index.faiss
- └── docstore.json
```
+索引文件取决于 Auto-Type:
+
+**AutoModel / AutoList**(自 v0.10.1 起为无 pickle 的 JSON):`index/index.json`。
+
+**图谱家族 / AutoSet / AutoDocument**(仍由 OMem 落盘):`index/index.faiss` + `index/docstore.json`。
+
---
## 语言支持
diff --git a/docs/zh/python/guides/saving-loading.md b/docs/zh/python/guides/saving-loading.md
index 8ae3f71..8a13b3c 100644
--- a/docs/zh/python/guides/saving-loading.md
+++ b/docs/zh/python/guides/saving-loading.md
@@ -29,6 +29,18 @@ result.dump("./my_ka/")
### 保存结构
+**AutoModel / AutoList**(自 v0.10.1 起为无 pickle 的 JSON):
+
+```
+./my_ka/
+├── data.json # 提取的知识
+├── metadata.json # 提取信息
+└── index/ # 搜索索引(如已构建)
+ └── index.json # 向量 + 文档,无 pickle
+```
+
+**图谱家族 / AutoSet / AutoDocument**(仍由 OMem 落盘,不是 v0.10.1 的 JSON 路径):
+
```
./my_ka/
├── data.json # 提取的知识
@@ -53,7 +65,7 @@ result.dump("./my_ka/")
## 加载知识库
!!! warning "只加载受信任的目录"
- 自 v0.10.1 起,索引以 JSON(`index.json`)存储,加载时不执行任何代码。旧版本(<= v0.10.0)的 KA 仍是 pickle 索引,加载时会反序列化——请只对自己生成或完全信任的知识库目录调用 `load()`(以及会加载 KA 的 CLI/MCP 命令)。用 `he build-index --force` 重建即可迁移旧 KA。
+ 请只对自己生成或完全信任的知识库目录调用 `load()`(以及会加载 KA 的 CLI/MCP 命令)。自 v0.10.1 起,**AutoModel / AutoList** 的索引以 JSON(`index.json`)存储,加载时不执行任何代码。旧版 model/list KA(<= v0.10.0)仍是 pickle 索引(`index.faiss` / `index.pkl`),加载时会反序列化——用 `he build-index --force` 重建即可迁移。图谱 / set / document 仍由 OMem 落盘,不走 v0.10.1 的 JSON 路径,因此这条信任警告仍然成立。
### 基本加载