feat: support Viking API key auth

config.yaml.full

@@ -168,8 +168,14 @@ database:
     # output_fields: text,metadata
   # [optional] for knowledgebase (https://console.volcengine.com/vikingdb)
   viking:
+    # api_key: ${DATABASE_VIKING_API_KEY}  # optional; for searching existing VikingDB knowledgebase collections
     project: default    # user project in Volcengine Viking DB
     region: cn-beijing
+  # [optional] for LongTermMemory(backend="viking")
+  vikingmem:
+    # api_key: ${DATABASE_VIKINGMEM_API_KEY}  # optional; for memory operations on existing VikingDB memory collections
+    project: default
+    memory_type: sys_event_v1,sys_profile_v1
   # [optional] for knowledgebase with viking database
   tos:
     endpoint: tos-cn-beijing.volces.com # default Volcengine TOS endpoint

docs/content/docs/framework/knowledgebase/overview.en.mdx

@@ -136,12 +136,14 @@ model:
     api_key: ${MODEL_AGENT_API_KEY}
 
 volcengine:
-  # Viking DB and embedding require Volcengine ak/sk
+  # Viking DB and embedding can use Volcengine ak/sk
   access_key: ${VOLCENGINE_ACCESS_KEY}
   secret_key: ${VOLCENGINE_SECRET_KEY}
 
 database:
   viking:
+    # Optional; used to search existing VikingDB knowledgebase collections
+    api_key: ${DATABASE_VIKING_API_KEY}
     project: default    # project in Volcengine Viking DB
     region: cn-beijing
   tos:
@@ -167,12 +169,15 @@ Provide those secrets in environment variables:
 
 ```bash
 export MODEL_AGENT_API_KEY="<your Ark API key>"
+export DATABASE_VIKING_API_KEY="<your VikingDB knowledgebase API key>"
 export VOLCENGINE_ACCESS_KEY="<your AK>"
 export VOLCENGINE_SECRET_KEY="<your SK>"
 export DATABASE_OPENSEARCH_PASSWORD="<your OpenSearch password>"
 export DATABASE_OPENVIKING_API_KEY="<your OpenViking API key>"
 ```
 
+`DATABASE_VIKING_API_KEY` is preferred for searching existing VikingDB knowledgebase collections. Creating, deleting, or listing collections, managing documents/chunks, and operations such as `add_from_text` / `add_from_files` that upload through TOS still require usable `VOLCENGINE_ACCESS_KEY` / `VOLCENGINE_SECRET_KEY` or VeFaaS IAM credentials.
+
 ### Demo scenario
 
 The example below uses short bios of two psychologists as knowledge, plus a date-difference tool, to demonstrate "retrieve from the knowledge base + call a tool."

docs/content/docs/framework/knowledgebase/overview.mdx

@@ -136,12 +136,14 @@ model:
     api_key: ${MODEL_AGENT_API_KEY}
 
 volcengine:
-  # Viking DB 与 embedding 需要火山引擎 ak/sk
+  # Viking DB 与 embedding 可使用火山引擎 ak/sk
   access_key: ${VOLCENGINE_ACCESS_KEY}
   secret_key: ${VOLCENGINE_SECRET_KEY}
 
 database:
   viking:
+    # 可选；用于搜索已有 VikingDB 知识库
+    api_key: ${DATABASE_VIKING_API_KEY}
     project: default    # Volcengine Viking DB 中的项目
     region: cn-beijing
   tos:
@@ -167,12 +169,15 @@ database:
 
 ```bash
 export MODEL_AGENT_API_KEY="<你的方舟 API Key>"
+export DATABASE_VIKING_API_KEY="<你的 VikingDB 知识库 API Key>"
 export VOLCENGINE_ACCESS_KEY="<你的 AK>"
 export VOLCENGINE_SECRET_KEY="<你的 SK>"
 export DATABASE_OPENSEARCH_PASSWORD="<你的 OpenSearch 密码>"
 export DATABASE_OPENVIKING_API_KEY="<你的 OpenViking API Key>"
 ```
 
+`DATABASE_VIKING_API_KEY` 会优先用于搜索已有 VikingDB 知识库。创建、删除、列举 collection，管理文档/切片，以及 `add_from_text` / `add_from_files` 这类需要上传 TOS 的能力，仍需要可用的 `VOLCENGINE_ACCESS_KEY` / `VOLCENGINE_SECRET_KEY` 或 VeFaaS IAM 凭证。
+
 ### 演示场景
 
 下面用一段心理学家简介作为知识，并配一个计算日期差的工具，演示「检索知识库 + 调用工具」。

docs/content/docs/framework/memory/long-term/vikingdb.en.mdx

@@ -11,10 +11,11 @@ The `viking` backend (`VikingDBLTMBackend`) uses Volcengine [VikingDB memory](ht
 
 ## Configuration
 
-Authenticated via Volcengine AK/SK. VeADK reads AK/SK from environment variables first, falling back to VeFaaS IAM credentials (for cloud deployments).
+Authenticate with a VikingDB memory API key or Volcengine AK/SK. VeADK prefers `api_key` / `DATABASE_VIKINGMEM_API_KEY` for memory operations on an existing collection; when it is unset, it uses AK/SK and then falls back to VeFaaS IAM credentials for cloud deployments. Collection creation, listing, and deletion still require AK/SK or IAM credentials.
 
 | Field | Env var | Default | Description |
 | :--- | :--- | :--- | :--- |
+| `api_key` | `DATABASE_VIKINGMEM_API_KEY` | — | VikingDB memory API key for memory operations on an existing collection. |
 | `volcengine_access_key` | `VOLCENGINE_ACCESS_KEY` | — | Volcengine Access Key. |
 | `volcengine_secret_key` | `VOLCENGINE_SECRET_KEY` | — | Volcengine Secret Key. |
 | `volcengine_project` | `DATABASE_VIKINGMEM_PROJECT` | `default` | VikingDB memory project. |

docs/content/docs/framework/memory/long-term/vikingdb.mdx

@@ -11,10 +11,11 @@ title: "VikingDB（viking）"
 
 ## 配置
 
-通过火山引擎 AK/SK 鉴权。VeADK 会优先从环境变量读取 AK/SK，未设置时尝试从 VeFaaS IAM 获取凭证（云上部署场景）。
+可以通过 VikingDB 记忆库 API Key 或火山引擎 AK/SK 鉴权。VeADK 会优先使用 `api_key` / `DATABASE_VIKINGMEM_API_KEY` 访问已有 collection 中的记忆数据；未设置时使用 AK/SK，仍未设置时尝试从 VeFaaS IAM 获取凭证（云上部署场景）。创建、列举、删除 collection 等管理操作仍需要 AK/SK 或 IAM 凭证。
 
 | 配置项 | 环境变量 | 默认值 | 说明 |
 | :--- | :--- | :--- | :--- |
+| `api_key` | `DATABASE_VIKINGMEM_API_KEY` | — | VikingDB 记忆库 API Key，用于访问已有 collection 中的记忆数据。 |
 | `volcengine_access_key` | `VOLCENGINE_ACCESS_KEY` | — | 火山引擎 Access Key。 |
 | `volcengine_secret_key` | `VOLCENGINE_SECRET_KEY` | — | 火山引擎 Secret Key。 |
 | `volcengine_project` | `DATABASE_VIKINGMEM_PROJECT` | `default` | VikingDB 记忆库项目。 |

docs/content/docs/references/configuration/environment-variables.en.mdx

@@ -108,8 +108,12 @@ Prefix `DATABASE_`, grouped by storage type. Memory and the knowledge base read
 | | `DATABASE_MILVUS_OVERWRITE` | Overwrite the collection, default `false` |
 | | `DATABASE_MILVUS_TIMEOUT` | Connection/request timeout (optional) |
 | | `DATABASE_MILVUS_OUTPUT_FIELDS` | Comma-separated retrieval output fields (optional) |
+| VikingDB knowledgebase | `DATABASE_VIKING_API_KEY` | API key for searching existing knowledgebases; management still requires AK/SK/IAM |
 | VikingDB | `DATABASE_VIKING_PROJECT` | Project name |
 | | `DATABASE_VIKING_REGION` | Region |
+| VikingDB memory | `DATABASE_VIKINGMEM_API_KEY` | API key for existing memory collections; management still requires AK/SK/IAM |
+| | `DATABASE_VIKINGMEM_PROJECT` | Project name, default `default` |
+| | `DATABASE_VIKINGMEM_MEMORY_TYPE` | Comma-separated memory types |
 | TOS | `DATABASE_TOS_ENDPOINT` | Endpoint |
 | | `DATABASE_TOS_REGION` | Region |
 | | `DATABASE_TOS_BUCKET` | Bucket name |

docs/content/docs/references/configuration/environment-variables.mdx

@@ -108,8 +108,12 @@ volcengine:
 | | `DATABASE_MILVUS_OVERWRITE` | 是否覆盖 collection，默认 `false` |
 | | `DATABASE_MILVUS_TIMEOUT` | 连接/请求超时时间（可选） |
 | | `DATABASE_MILVUS_OUTPUT_FIELDS` | 检索输出字段，逗号分隔（可选） |
+| VikingDB 知识库 | `DATABASE_VIKING_API_KEY` | API Key，用于搜索已有知识库；管理操作仍需 AK/SK/IAM |
 | VikingDB | `DATABASE_VIKING_PROJECT` | 项目名称 |
 | | `DATABASE_VIKING_REGION` | 区域 |
+| VikingDB 记忆库 | `DATABASE_VIKINGMEM_API_KEY` | API Key，用于访问已有记忆 collection；管理操作仍需 AK/SK/IAM |
+| | `DATABASE_VIKINGMEM_PROJECT` | 项目名称，默认 `default` |
+| | `DATABASE_VIKINGMEM_MEMORY_TYPE` | 记忆类型，逗号分隔 |
 | TOS | `DATABASE_TOS_ENDPOINT` | 访问端点 |
 | | `DATABASE_TOS_REGION` | 区域 |
 | | `DATABASE_TOS_BUCKET` | 桶名称 |

pyproject.toml

@@ -52,7 +52,7 @@ dependencies = [
     "pypdfium2>=4.30.0", # Frontend PDF attachments are rendered as page images
     "pillow>=10.0.0", # Encode rendered PDF pages as PNG
     "qrcode>=8,<9", # Render Feishu app-registration links for Studio setup
-    "vikingdb-python-sdk>=0.1.3", # For Viking DB
+    "vikingdb-python-sdk>=0.1.32", # For Viking DB
     "agentkit-sdk-python>=0.8.0",
     "websockets>=15,<16", # Stream temporary AgentKit Sandbox conversations
     "openviking-sdk>=0.1.3",