refactor(executor, handlers): replace channel-based streams with StreamResult for consistency

- Updated `ExecuteStream` functions in executors to use `StreamResult` instead of channels.
- Enhanced upstream header handling in OpenAI handlers.
- Improved maintainability and alignment across executors and handlers.

internal/runtime/executor/github_copilot_executor.go

@@ -35,12 +35,12 @@ const (
 	maxScannerBufferSize = 20_971_520
 
 	// Copilot API header values.
-	copilotUserAgent      = "GitHubCopilotChat/0.35.0"
-	copilotEditorVersion  = "vscode/1.107.0"
-	copilotPluginVersion  = "copilot-chat/0.35.0"
-	copilotIntegrationID  = "vscode-chat"
-	copilotOpenAIIntent   = "conversation-panel"
-	copilotGitHubAPIVer   = "2025-04-01"
+	copilotUserAgent     = "GitHubCopilotChat/0.35.0"
+	copilotEditorVersion = "vscode/1.107.0"
+	copilotPluginVersion = "copilot-chat/0.35.0"
+	copilotIntegrationID = "vscode-chat"
+	copilotOpenAIIntent  = "conversation-panel"
+	copilotGitHubAPIVer  = "2025-04-01"
 )
 
 // GitHubCopilotExecutor handles requests to the GitHub Copilot API.
@@ -232,7 +232,7 @@ func (e *GitHubCopilotExecutor) Execute(ctx context.Context, auth *cliproxyauth.
 }
 
 // ExecuteStream handles streaming requests to GitHub Copilot.
-func (e *GitHubCopilotExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
+func (e *GitHubCopilotExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (_ *cliproxyexecutor.StreamResult, err error) {
 	apiToken, baseURL, errToken := e.ensureAPIToken(ctx, auth)
 	if errToken != nil {
 		return nil, errToken
@@ -341,7 +341,6 @@ func (e *GitHubCopilotExecutor) ExecuteStream(ctx context.Context, auth *cliprox
 	}
 
 	out := make(chan cliproxyexecutor.StreamChunk)
-	stream = out
 
 	go func() {
 		defer close(out)
@@ -394,7 +393,10 @@ func (e *GitHubCopilotExecutor) ExecuteStream(ctx context.Context, auth *cliprox
 		}
 	}()
 
-	return stream, nil
+	return &cliproxyexecutor.StreamResult{
+		Headers: httpResp.Header.Clone(),
+		Chunks:  out,
+	}, nil
 }
 
 // CountTokens is not supported for GitHub Copilot.

internal/runtime/executor/kilo_executor.go

@@ -168,7 +168,7 @@ func (e *KiloExecutor) Execute(ctx context.Context, auth *cliproxyauth.Auth, req
 }
 
 // ExecuteStream performs a streaming request.
-func (e *KiloExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
+func (e *KiloExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (_ *cliproxyexecutor.StreamResult, err error) {
 	baseModel := thinking.ParseSuffix(req.Model).ModelName
 
 	reporter := newUsageReporter(ctx, e.Identifier(), baseModel, auth)
@@ -253,7 +253,6 @@ func (e *KiloExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
 	}
 
 	out := make(chan cliproxyexecutor.StreamChunk)
-	stream = out
 	go func() {
 		defer close(out)
 		defer httpResp.Body.Close()
@@ -286,7 +285,10 @@ func (e *KiloExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
 		reporter.ensurePublished(ctx)
 	}()
 
-	return stream, nil
+	return &cliproxyexecutor.StreamResult{
+		Headers: httpResp.Header.Clone(),
+		Chunks:  out,
+	}, nil
 }
 
 // Refresh validates the Kilo token.
@@ -456,4 +458,3 @@ func FetchKiloModels(ctx context.Context, auth *cliproxyauth.Auth, cfg *config.C
 
 	return allModels
 }
-

internal/runtime/executor/kiro_executor.go

@@ -1053,7 +1053,7 @@ func (e *KiroExecutor) executeWithRetry(ctx context.Context, auth *cliproxyauth.
 
 // ExecuteStream handles streaming requests to Kiro API.
 // Supports automatic token refresh on 401/403 errors and quota fallback on 429.
-func (e *KiroExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (stream <-chan cliproxyexecutor.StreamChunk, err error) {
+func (e *KiroExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Auth, req cliproxyexecutor.Request, opts cliproxyexecutor.Options) (_ *cliproxyexecutor.StreamResult, err error) {
 	accessToken, profileArn := kiroCredentials(auth)
 	if accessToken == "" {
 		return nil, fmt.Errorf("kiro: access token not found in auth")
@@ -1110,7 +1110,11 @@ func (e *KiroExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
 	// Route to MCP endpoint instead of normal Kiro API
 	if kiroclaude.HasWebSearchTool(req.Payload) {
 		log.Infof("kiro: detected pure web_search request, routing to MCP endpoint")
-		return e.handleWebSearchStream(ctx, auth, req, opts, accessToken, profileArn)
+		streamWebSearch, errWebSearch := e.handleWebSearchStream(ctx, auth, req, opts, accessToken, profileArn)
+		if errWebSearch != nil {
+			return nil, errWebSearch
+		}
+		return &cliproxyexecutor.StreamResult{Chunks: streamWebSearch}, nil
 	}
 
 	reporter := newUsageReporter(ctx, e.Identifier(), req.Model, auth)
@@ -1128,7 +1132,11 @@ func (e *KiroExecutor) ExecuteStream(ctx context.Context, auth *cliproxyauth.Aut
 
 	// Execute stream with retry on 401/403 and 429 (quota exhausted)
 	// Note: currentOrigin and kiroPayload are built inside executeStreamWithRetry for each endpoint
-	return e.executeStreamWithRetry(ctx, auth, req, opts, accessToken, effectiveProfileArn, nil, body, from, reporter, "", kiroModelID, isAgentic, isChatOnly, tokenKey)
+	streamKiro, errStreamKiro := e.executeStreamWithRetry(ctx, auth, req, opts, accessToken, effectiveProfileArn, nil, body, from, reporter, "", kiroModelID, isAgentic, isChatOnly, tokenKey)
+	if errStreamKiro != nil {
+		return nil, errStreamKiro
+	}
+	return &cliproxyexecutor.StreamResult{Chunks: streamKiro}, nil
 }
 
 // executeStreamWithRetry performs the streaming HTTP request with automatic retry on auth errors.