docs(kitex): add link to Kitex v0.13.1 release notes

docs: create generic streaming documentation in Chinese version (#1326)

docs: 明确NewOpenTelemetryProvider的重复调用规范（关联Issue #1327） (#1328)

修复milvus的indexer文档，缺乏”document配置必须搭配fields配置“这一信息，导致容易误用问题 (#1331)

chore: upgrade typos (#1332)

feat: update merge register & with eager execution (#1333)

chore: update getting-started docs (#1334)

feat: update host multi-agent docs (#1335)

docs: add doc for hertz v0.10.0 (#1337)

Co-authored-by: Kyle Xiao <xiaoguoqiao@bytedance.com>

fix(eino): remove error prompt directory (#1338)

docs(kitex): improve transport protocol documentation with clearer explanations

docs: improve transport protocol documentation to clarify Kitex's support for Thrift and Protobuf with gRPC transport

Author: XinyiWang0204

.github/workflows/push-check.yml

@@ -9,4 +9,4 @@ jobs:
       - uses: actions/checkout@v4
 
       - name: typos-action
-        uses: crate-ci/typos@v1.14.3
+        uses: crate-ci/typos@v1.32.0

_typos.toml

@@ -3,8 +3,14 @@
 [files]
 extend-exclude = ["*.json", "*.js", "static/*", "layouts/*", "assets/*", "i18n/*"]
 
+[default]
+extend-ignore-re = [
+    "mmaping files"
+]
+
 [default.extend-words]
 datas = "datas"
 mmaped = "mmaped"
 exten = "exten"
-invokable = "invokable"
+invokable = "invokable"
+typ = "typ"

content/en/blog/releases/Hertz/release-v0_10_0.md

@@ -0,0 +1,36 @@
+---
+title: "Hertz Release v0.10.0"
+linkTitle: "Release v0.10.0"
+projects: ["Hertz"]
+date: 2025-05-21
+description: >
+---
+
+The Hertz v0.10.0 release adds two features and some fixes.
+
+1. Integrate SSE functionality. Refer to [SSE](/docs/hertz/tutorials/basic-feature/sse) for usage.
+2. Added http.Handler adaptor, extending Hertz using the official net/http ecosystem. Refer to [Adaptor](/docs/hertz/tutorials/basic-feature/http-adaptor) for usage.
+
+## Feature
+1. [[#1327](https://github.com/cloudwego/hertz/pull/1327)] feat(adaptor): new HertzHandler for http.Handler
+2. [[#1349](https://github.com/cloudwego/hertz/pull/1349)] feat(sse): SetLastEventID
+3. [[#1343](https://github.com/cloudwego/hertz/pull/1343)] feat(sse): reader supports cancel stream
+4. [[#1341](https://github.com/cloudwego/hertz/pull/1341)] feat(server): detect request race
+5. [[#1339](https://github.com/cloudwego/hertz/pull/1339)] feat(sse): add LastEventID helpers
+6. [[#1335](https://github.com/cloudwego/hertz/pull/1335)] feat(protocol): new sse pkg
+7. [[#1322](https://github.com/cloudwego/hertz/pull/1322)] feat: std transport sense client connection close
+
+## Fix
+1. [[#1340](https://github.com/cloudwego/hertz/pull/1340)] fix: only use netpoll & sonic on amd64/arm64 linux/darwin
+2. [[#1333](https://github.com/cloudwego/hertz/pull/1333)] fix(protocol): unexpected set resp.bodyStream
+3. [[#1329](https://github.com/cloudwego/hertz/pull/1329)] fix(client): stream body for sse instead of timeout
+4. [[#1332](https://github.com/cloudwego/hertz/pull/1332)] fix(server): Shutdown checks ExitWaitTimeout
+5. [[#1316](https://github.com/cloudwego/hertz/pull/1316)] fix: prioritize custom validator
+
+## Tests
+1. [[#1336](https://github.com/cloudwego/hertz/pull/1336)] test(protocol): fix hardcoded listen addr
+
+## Chore
+1. [[#1353](https://github.com/cloudwego/hertz/pull/1353)] chore: update netpoll dependency
+2. [[#1337](https://github.com/cloudwego/hertz/pull/1337)] chore(hz): update hz tool v0.9.7
+3. [[#1328](https://github.com/cloudwego/hertz/pull/1328)] ci: disable codecov annotations

content/en/blog/releases/Kitex/release-v043.md

@@ -18,7 +18,7 @@ description: >
 ### **Optimization**
 
 1. **Frugal Performance Optimization** : Support frugal precompile (pretouch) when new client or server, which is to reduce the impact of dynamic compilation on latency.
-2. **Connpool Optimiztion** : Refactor connection pool to improve the idle connections cleanup.
+2. **Connpool Optimization** : Refactor connection pool to improve the idle connections cleanup.
 
 ---
 

content/en/blog/releases/Kitex/release-v0_13_0.md

@@ -93,7 +93,7 @@ description: >
 
 [[#1725](https://github.com/cloudwego/kitex/pull/1725)] fix: code gen import issue for streamx mode, stream call judgement bug and set ttheader streaming as default
 
-[[#1727](https://github.com/cloudwego/kitex/pull/1727)] fix(tool): fix tool UseStdLib remains unexcepted lib issue
+[[#1727](https://github.com/cloudwego/kitex/pull/1727)] fix(tool): fix tool UseStdLib remains unexpected lib issue
 
 ### Refactor
 [[#1658](https://github.com/cloudwego/kitex/pull/1658)] refactor: streamx api to adapt both grpc and ttheader streaming protocol and provide more user-friendly interface

content/en/cooperation/shumei.md

@@ -250,7 +250,7 @@ Kitex 提供了一个很好的机制，叫**重试熔断**的机制，其中有
 
 **Thrift** **（C++）/ Fbthrift 服务的过载保护**
 
-> 基础服务层大多是计算密集型服务，CPU 利用率等不完全反应程序的负载，服务过载可能是 GPU 负载过高。 使用请求在队列中的平均等待时间（“queue time”）来计算服务的负载。之所以不选择请求的平均处理时间（“procss time”），是为了去除下游服务调用的影响。有时 process time 的增加并不代表当前服务过载了，而是请求依赖的下游服务过载。另一方面，请求的 process time 增加到一定程度，当前服务的资源也会逐渐耗尽，最终反应在 queue time 的增加上。
+> 基础服务层大多是计算密集型服务，CPU 利用率等不完全反应程序的负载，服务过载可能是 GPU 负载过高。 使用请求在队列中的平均等待时间（“queue time”）来计算服务的负载。之所以不选择请求的平均处理时间（“process time”），是为了去除下游服务调用的影响。有时 process time 的增加并不代表当前服务过载了，而是请求依赖的下游服务过载。另一方面，请求的 process time 增加到一定程度，当前服务的资源也会逐渐耗尽，最终反应在 queue time 的增加上。
 
 我们可以看下图，当我们整个 CPU 或者是 GPU 整体的负载较高的时候，它的排队时间就会去无限长。所以这里给我们一种启发的机制，也就是说我们设置一个最长的队列等待时间。因为刚才也说了，在整个微服务的框架下，我们一定要干的一件事情就是一定要有超时的机制，因为没有超时的机制最终可能导致整个系统都被焊住。所以我们这里针对于这个 Queue Time 设置一个 Timeout 的话，那就可以去保证我们这些计算密集能服务在它的排队时间，等到一定时间的时候我们就立即丢弃请求。
 

content/en/docs/eino/core_modules/chain_and_graph_orchestration/orchestration_design_principles.md

@@ -359,6 +359,13 @@ graph.AddLambdaNode("your_node_key", compose.InvokableLambda(func(ctx context.Co
 }), compose.WithOutputKey("your_output_key"))
 ```
 
+You can also register custom merge method to support any type:
+
+```go
+// eino/compose/values_merge.go
+func RegisterValuesMergeFunc[T any](fn func([]T) (T, error))
+```
+
 Workflow can map the output fields of multiple predecessor nodes to different input fields of the successor node. Eino converts the Struct output from each predecessor to a Map before any merge process, still conforming to the above rules.
 
 ### **Streaming Processing**
@@ -464,4 +471,9 @@ When `NodeTriggerMode == AllPredecessor`, the graph executes using the dag engin
 - Support Branch. At runtime, mark the unselected nodes of Branch as skipped, which does not affect the semantics of AllPredecessor.
 - No need for manual SuperStep alignment.
 
+> 💡
+> When NodeTriggerMode is set to AllPredecessor, the node will execute after all predecessors are ready, but it will not execute immediately. Instead, it still follows the SuperStep - running new runnable nodes after a batch of nodes have completed execution.
+>
+> If compose.WithEagerExecution() is passed in Compile, the ready nodes will run immediately.
+
 In summary, the pregel mode is flexible and powerful but comes with additional mental overhead, while the dag mode is clear and simple but limited in application scenarios. In the Eino framework, Chain uses the pregel mode, Workflow uses the dag mode, and Graph supports both; users can choose between pregel and dag.

content/en/docs/eino/core_modules/flow_integration_components/multi_agent_hosting.md

@@ -343,3 +343,77 @@ HandOff to answer_with_journal with argument {"reason":"To find out the user's m
 Answer:
 You got up at 7:00 in the morning.
 ```
+
+## FAQ
+
+### Host direct answer does not have streaming effect
+
+Host Multi-Agent provides a configuration for `StreamToolCallChecker` to determine whether the Host outputs directly.
+
+Different models may output tool calls in different ways in streaming mode: some models (e.g., OpenAI) output tool calls directly; some models (e.g., Claude) output text first and then output tool calls. Therefore, different methods are needed for the determination. This field is used to specify a function for determining whether the model's streaming output contains tool calls.
+
+It is optional. If not filled, the determination of whether the "non-empty package" contains tool calls is used:
+
+```go
+func firstChunkStreamToolCallChecker(_ context.Context, sr *schema.StreamReader[*schema.Message]) (bool, error) {
+    defer sr.Close()
+
+    for {
+       msg, err := sr.Recv()
+       if err == io.EOF {
+          return false, nil
+       }
+       if err != nil {
+          return false, err
+       }
+
+       if len(msg.ToolCalls) > 0 {
+          return true, nil
+       }
+
+       if len(msg.Content) == 0 { // skip empty chunks at the front
+          continue
+       }
+
+       return false, nil
+    }
+}
+```
+
+The above default implementation is applicable when the Tool Call Message output by the model contains only Tool Calls.
+
+The default implementation is not applicable when there is a non-empty content chunk before outputting the Tool Call. In this case, a custom tool call checker is required as follows:
+
+```go
+toolCallChecker := func(ctx context.Context, sr *schema.StreamReader[*schema.Message]) (bool, error) {
+    defer sr.Close()
+    for {
+       msg, err := sr.Recv()
+       if err != nil {
+          if errors.Is(err, io.EOF) {
+             // finish
+             break
+          }
+
+          return false, err
+       }
+
+       if len(msg.ToolCalls) > 0 {
+          return true, nil
+       }
+    }
+    return false, nil
+}
+```
+
+This custom `StreamToolCallChecker` may need to check **all packages** for the presence of ToolCalls in extreme cases, resulting in the loss of the "streaming judgment" effect. If you want to maintain the "streaming judgment" effect as much as possible, the suggested solution is:
+
+Try to add a prompt to constrain the model not to output additional text when making tool calls, for example: "If you need to call a tool, output the tool directly without outputting text."
+
+Different models may be affected by the prompt differently. You need to adjust the prompt and verify the effect in actual use.
+
+### Host picks multiple Specialists simultaneously
+
+The Host provides the selection of Specialists in the form of Tool Calls, so it may select multiple Specialists simultaneously in the form of a Tool Call list. At this time, the Host Multi-Agent will route the request to these multiple Specialists at the same time. After the multiple Specialists complete their tasks, the Summarizer node will summarize multiple Messages into one Message as the final output of the Host Multi-Agent.
+
+Users can customize the behavior of the Summarizer by configuring the Summarizer, specifying a ChatModel and a SystemPrompt. If not specified, the Host Multi-Agent will concatenate the output Message Contents of multiple Specialists and return the result.

content/en/docs/eino/ecosystem_integration/indexer/indexer_milvus.md

@@ -116,6 +116,7 @@ type IndexerConfig struct {
     // Default Collection config
     // Collection is the collection name in milvus database
     // Optional, and the default value is "eino_collection"
+    // If you want to use this configuration, you must include the Fields configuration
     Collection string
     // Description is the description for collection
     // Optional, and the default value is "the collection for eino"

content/en/docs/eino/ecosystem_integration/prompt/prompt_mcp.md

@@ -45,3 +45,7 @@ In this example, we will use a react agent to build an agent that can interact w
 
 - Understand the core modules and concepts of Eino: [Eino: Core Modules](/docs/eino/core_modules), which is crucial information for mastering the use of Eino for application development.
 - Eino maintains an open ecosystem stance and provides a large number of ecosystem integration components: [Eino: ecosystem](/docs/eino/ecosystem). You can use these components to quickly build your own business applications.
+
+
+If you find this project is good and helpful, please give [Eino](https://github.com/cloudwego/eino) a star! Your support will be our greatest encouragement!  
+If you have any questions or suggestions, feel free to leave a message in [GitHub Issues](https://github.com/cloudwego/eino/issues).

content/en/docs/eino/quick_start/_index.md

@@ -186,3 +186,6 @@ hz update -idl hello.thrift
 ## More examples
 
 Please refer：[Example code](/docs/hertz/tutorials/example/)
+
+If you think this project is good and helpful, please give [Hertz](https://github.com/cloudwego/hertz) a star! This will be the greatest encouragement and support for us!
+If you have any questions or suggestions, feel free to leave a message in [GitHub Issues](https://github.com/cloudwego/hertz/issues) according to the issue template.

content/en/docs/hertz/getting-started/_index.md

@@ -23,6 +23,8 @@ keywords:
     "Constants",
     "Render",
     "JSON Marshal Library",
+    "SSE",
+    "http.Handler adaptor",
   ]
 description: "Basic feature of Hertz."
 ---

content/en/docs/hertz/tutorials/basic-feature/_index.md

@@ -375,14 +375,14 @@ Use RequestContext.Response.Header to obtain the ResponseHeader object, this obj
 | `func (h *ResponseHeader) DelClientCookieBytes(key []byte)`                     | Remove the cookie from the client                                                                                                                                                                                                |
 | `func (h *ResponseHeader) SetConnectionClose(close bool)`                       | Set the `Connection: close` flag in the response header                                                                                                                                                                          |
 | `func (h *ResponseHeader) ConnectionClose() bool`                               | Determine if Connection: close is included                                                                                                                                                                                       |
-| `func (h *ResponseHeader) ContentEncoding() []byte`                             | Obtion Content-Encoding                                                                                                                                                                                                          |
+| `func (h *ResponseHeader) ContentEncoding() []byte`                             | Option Content-Encoding                                                                                                                                                                                                          |
 | `func (h *ResponseHeader) SetContentEncoding(contentEncoding string)`           | Set Content-Encoding                                                                                                                                                                                                             |
 | `func (h *ResponseHeader) SetContentEncodingBytes(contentEncoding []byte)`      | Set Content-Encoding                                                                                                                                                                                                             |
 | `func (h *ResponseHeader) SetCanonical(key, value []byte)`                      | Set the Header key value, assuming that the key is in canonical form                                                                                                                                                             |
 | `func (h *ResponseHeader) Server() []byte`                                      | Return the value with key `Server` in the header                                                                                                                                                                                 |
 | `func (h *ResponseHeader) SetServerBytes(server []byte)`                        | Set the key in the header to the value of Server                                                                                                                                                                                 |
 | `func (h *ResponseHeader) MustSkipContentLength() bool`                         | Determine if there is a response body (according to the HTTP/1.1 protocol, there is no response body when the response status codes are 1xx, 204, or 304)                                                                        |
-| `func (h *ResponseHeader) StatusCode() int`                                     | Obtion StatusCode                                                                                                                                                                                                                |
+| `func (h *ResponseHeader) StatusCode() int`                                     | Option StatusCode                                                                                                                                                                                                                |
 | `func (h *ResponseHeader) SetStatusCode(statusCode int)`                        | Set StatusCode                                                                                                                                                                                                                   |
 | `func (h *ResponseHeader) Len() int`                                            | Return the number of headers                                                                                                                                                                                                     |
 | `func (h *ResponseHeader) DisableNormalizing()`                                 | Disable the normalization of header name (capitalize the first letter and the first letter after the Em dash)                                                                                                                    |

content/en/docs/hertz/tutorials/basic-feature/context/response.md

@@ -0,0 +1,50 @@
+---
+title: "http.Handler adaptor"
+date: 2025-05-21
+weight: 20
+keywords: ["adaptor"]
+description: ""
+---
+
+## Background
+
+- Phasing out the poorly maintained hertz app.FS implementation
+- Providing a compatible and performant fs implementation through the official net/http ecosystem
+- Extending hertz functionality through the official net/http ecosystem to reduce custom implementations
+
+## What is adaptor.HertzHandler
+
+- Allows you to convert existing http.HandlerFunc methods directly to HertzHandler
+- Enables direct use of standard library methods like http.FileServer and embed.FS
+- Even allows direct use of github.com/gorilla/websocket in Hertz
+
+## Example
+
+```go
+package main
+
+import (
+  "embed"
+  "net/http"
+
+  "github.com/cloudwego/hertz/pkg/app/server"
+  "github.com/cloudwego/hertz/pkg/common/adaptor"
+)
+
+//go:embed static/*
+var staticFiles embed.FS
+
+func main() {
+  h := server.Default()
+
+  helloHandler := http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+    w.Write([]byte("Hello hertz!"))
+  })
+  h.GET("/hello", adaptor.HertzHandler(helloHandler))
+
+  staticFS := adaptor.HertzHandler(http.FileServer(http.FS(staticFiles)))
+  h.GET("/static/*filepath", staticFS)
+  h.HEAD("/static/*filepath", staticFS)
+  h.Spin()
+}
+```