feat: implement streaming chat, skill routing, and SAFe PI planning tools

- Add /api/chat/stream endpoint with Server-Sent Events (SSE) for real-time message streaming
  * Implement StreamEvent types (thought, tool_call, tool_result, final, error)
  * Add StreamEventCallback mechanism for event propagation
  * Create StreamChatHandler in webui/bot with proper HTTP headers and flushing

- Implement LLM-based skill router for intelligent capability selection
  * Add optional routerLLM client for semantic routing
  * Implement routeSkillsWithLLM() to match user intent to available skills
  * Add matchSkillsByName() for fuzzy skill matching
  * Update buildUnifiedSystemPrompt() to use routed skills

- Add streaming support to ReAct pipeline
  * Implement runUnifiedReActStream() for streaming thought/action/observation
  * Emit StreamEvent at each ReAct step
  * Support callback error handling in streaming mode

- Integrate three new DevOps tools
  * tools/filedoc: Extract document content from file_id via OpenAI
  * tools/giteaticket: Create Gitea issues from PI plan items with SAFe metadata
  * tools/piplan: Publish PI planning blueprints with dependency tracking

- Add SAFe PI Planning skill
  * Implement PM/SA/RTE (iron triangle) workflow
  * Support for Feature, Enabler, and Dependency definition
  * Automatic task decomposition and Gitea integration

- Create frontend integration documentation
  * Complete SSE protocol specification
  * TypeScript fetch + ReadableStream example
  * LLM-ready refactoring template for other projects

- Simplify file handling
  * Remove legacy file context structures and dual-mode processing
  * Consolidate file operations into UploadAndCacheFiles()
  * Remove FilePromptMode configuration and related complexity

- Update configuration
  * Add Router model support (LLM_ROUTER_MODEL)
  * Add Gitea configuration (BaseURL, Token, Owner, Repo)
  * WebSearch and additional tool infrastructure

Tests: All 22 test packages passing, 8/8 webui tests including 3 new stream tests

internal/llm/client.go

@@ -33,21 +33,13 @@ type MessageChatClient interface {
 	GenerateMessages(ctx context.Context, messages []PromptMessage) (string, error)
 }
 
-type FileChatClient interface {
-	GenerateWithFiles(ctx context.Context, systemPrompt, userPrompt string, fileIDs []string, appendFileIDText bool) (string, error)
-}
-
-type FileMessageChatClient interface {
-	GenerateMessagesWithFiles(ctx context.Context, messages []PromptMessage, fileIDs []string, appendFileIDText bool) (string, error)
-}
-
 type FileUploader interface {
 	UploadFile(ctx context.Context, file InputFile, purpose string) (string, error)
 }
 
 // ToolCallChatClient 支持原生 function calling 的 LLM 客户端接口。
 type ToolCallChatClient interface {
-	GenerateWithTools(ctx context.Context, messages []PromptMessage, tools []ToolDefinition, fileIDs []string, appendFileIDText bool) (*ChatCompletion, error)
+	GenerateWithTools(ctx context.Context, messages []PromptMessage, tools []ToolDefinition) (*ChatCompletion, error)
 }
 
 // ToolDefinition 描述一个可供 LLM 调用的工具函数定义。
@@ -89,11 +81,9 @@ type InputFile struct {
 }
 
 type OpenAICompatibleClient struct {
-	client         openai.Client
-	model          string
-	fileModel      string
-	filePromptMode string
-	log            *logger.Logger
+	client openai.Client
+	model  string
+	log    *logger.Logger
 }
 
 func NewOpenAICompatibleClient(cfg config.LLMConfig, log *logger.Logger) *OpenAICompatibleClient {
@@ -105,11 +95,9 @@ func NewOpenAICompatibleClient(cfg config.LLMConfig, log *logger.Logger) *OpenAI
 		opts = append(opts, option.WithBaseURL(cfg.BaseURL))
 	}
 	return &OpenAICompatibleClient{
-		client:         openai.NewClient(opts...),
-		model:          cfg.Model,
-		fileModel:      cfg.FileModel,
-		filePromptMode: cfg.FilePromptMode,
-		log:            log,
+		client: openai.NewClient(opts...),
+		model:  cfg.Model,
+		log:    log,
 	}
 }
 
@@ -118,38 +106,22 @@ func (c *OpenAICompatibleClient) Generate(ctx context.Context, systemPrompt, use
 		{Role: "system", Content: systemPrompt},
 		{Role: "user", Content: userPrompt},
 	}
-	return c.generateWithMessagesInternal(ctx, messages, nil, false)
-}
-
-func (c *OpenAICompatibleClient) GenerateWithFiles(ctx context.Context, systemPrompt, userPrompt string, fileIDs []string, appendFileIDText bool) (string, error) {
-	messages := []PromptMessage{
-		{Role: "system", Content: systemPrompt},
-		{Role: "user", Content: userPrompt},
-	}
-	return c.generateWithMessagesInternal(ctx, messages, fileIDs, appendFileIDText)
+	return c.generateWithMessagesInternal(ctx, messages)
 }
 
 func (c *OpenAICompatibleClient) GenerateMessages(ctx context.Context, messages []PromptMessage) (string, error) {
-	return c.generateWithMessagesInternal(ctx, messages, nil, false)
-}
-
-func (c *OpenAICompatibleClient) GenerateMessagesWithFiles(ctx context.Context, messages []PromptMessage, fileIDs []string, appendFileIDText bool) (string, error) {
-	return c.generateWithMessagesInternal(ctx, messages, fileIDs, appendFileIDText)
+	return c.generateWithMessagesInternal(ctx, messages)
 }
 
 // GenerateWithTools 使用原生 function calling 发送请求，返回结构化的 ChatCompletion。
-func (c *OpenAICompatibleClient) GenerateWithTools(ctx context.Context, messages []PromptMessage, tools []ToolDefinition, fileIDs []string, appendFileIDText bool) (*ChatCompletion, error) {
+func (c *OpenAICompatibleClient) GenerateWithTools(ctx context.Context, messages []PromptMessage, tools []ToolDefinition) (*ChatCompletion, error) {
 	model := c.model
-	ids := nonEmptyIDs(fileIDs)
-	if len(ids) > 0 && strings.TrimSpace(c.fileModel) != "" {
-		model = c.fileModel
-	}
 
-	sdkMessages := buildSDKMessages(messages, ids, c.normalizedFilePromptMode(), appendFileIDText)
+	sdkMessages := buildSDKMessages(messages)
 	sdkTools := toSDKTools(tools)
 
 	if c.log != nil {
-		c.log.Debugf("llm tool-call request start model=%s messages=%d tools=%d files=%d", model, len(sdkMessages), len(sdkTools), len(ids))
+		c.log.Debugf("llm tool-call request start model=%s messages=%d tools=%d", model, len(sdkMessages), len(sdkTools))
 	}
 
 	params := openai.ChatCompletionNewParams{
@@ -188,12 +160,8 @@ func (c *OpenAICompatibleClient) GenerateWithTools(ctx context.Context, messages
 	}, nil
 }
 
-func (c *OpenAICompatibleClient) generateWithMessagesInternal(ctx context.Context, messages []PromptMessage, fileIDs []string, appendFileIDText bool) (string, error) {
+func (c *OpenAICompatibleClient) generateWithMessagesInternal(ctx context.Context, messages []PromptMessage) (string, error) {
 	model := c.model
-	ids := nonEmptyIDs(fileIDs)
-	if len(ids) > 0 && strings.TrimSpace(c.fileModel) != "" {
-		model = c.fileModel
-	}
 
 	baseMessages := normalizePromptMessages(messages)
 	if len(baseMessages) == 0 {
@@ -202,10 +170,10 @@ func (c *OpenAICompatibleClient) generateWithMessagesInternal(ctx context.Contex
 
 	systemLen, userLen := promptMessageLengths(baseMessages)
 	if c.log != nil {
-		c.log.Debugf("llm request start model=%s system_len=%d user_len=%d file_count=%d file_prompt_mode=%s", model, systemLen, userLen, len(ids), c.normalizedFilePromptMode())
+		c.log.Debugf("llm request start model=%s system_len=%d user_len=%d", model, systemLen, userLen)
 	}
 
-	sdkMessages := buildSDKMessages(baseMessages, ids, c.normalizedFilePromptMode(), appendFileIDText)
+	sdkMessages := buildSDKMessages(baseMessages)
 
 	params := openai.ChatCompletionNewParams{
 		Model:    shared.ChatModel(model),
@@ -234,10 +202,9 @@ func (c *OpenAICompatibleClient) generateWithMessagesInternal(ctx context.Contex
 	return content, nil
 }
 
-// buildSDKMessages 将 PromptMessage 列表转换为 openai SDK 的消息格式，并注入 file_id（如需要）。
-func buildSDKMessages(base []PromptMessage, fileIDs []string, mode string, appendFileIDText bool) []openai.ChatCompletionMessageParamUnion {
-	mode = strings.ToLower(strings.TrimSpace(mode))
-	out := make([]openai.ChatCompletionMessageParamUnion, 0, len(base)+2)
+// buildSDKMessages 将 PromptMessage 列表转换为 openai SDK 的消息格式。
+func buildSDKMessages(base []PromptMessage) []openai.ChatCompletionMessageParamUnion {
+	out := make([]openai.ChatCompletionMessageParamUnion, 0, len(base))
 
 	for _, m := range base {
 		role := normalizeRole(m.Role)
@@ -247,34 +214,6 @@ func buildSDKMessages(base []PromptMessage, fileIDs []string, mode string, appen
 		out = append(out, toSDKMessage(m, role))
 	}
 
-	if len(fileIDs) == 0 {
-		return out
-	}
-
-	if appendFileIDText {
-		// WebUI 场景：将首个 fileID 作为 text part 追加到最后一个 user 消息。
-		firstFileID := strings.TrimSpace(fileIDs[0])
-		if firstFileID == "" {
-			return out
-		}
-		for i := len(out) - 1; i >= 0; i-- {
-			if r := out[i].GetRole(); r != nil && *r == "user" {
-				out[i] = buildUserMessageWithFileIDText(out[i], firstFileID)
-				return out
-			}
-		}
-		out = append(out, buildUserMessageWithFileIDText(openai.UserMessage(""), firstFileID))
-		return out
-	}
-
-	// 非 WebUI 场景：保持原有 file content part 方式。
-	for i := len(out) - 1; i >= 0; i-- {
-		if r := out[i].GetRole(); r != nil && *r == "user" {
-			out[i] = buildUserMessageWithFiles(out[i], fileIDs)
-			return out
-		}
-	}
-	out = append(out, buildUserMessageWithFiles(openai.UserMessage(""), fileIDs))
 	return out
 }
 
@@ -309,53 +248,6 @@ func toSDKMessage(m PromptMessage, role string) openai.ChatCompletionMessagePara
 	}
 }
 
-// buildUserMessageWithFileIDText 为 user 消息追加一个 text part，内容为 fileID。
-func buildUserMessageWithFileIDText(msg openai.ChatCompletionMessageParamUnion, fileID string) openai.ChatCompletionMessageParamUnion {
-	// 提取已有的文本内容
-	text := ""
-	if s, ok := msg.GetContent().AsAny().(*string); ok && s != nil {
-		text = *s
-	}
-	fileID = strings.TrimSpace(fileID)
-	if fileID == "" {
-		return msg
-	}
-
-	parts := make([]openai.ChatCompletionContentPartUnionParam, 0, 2)
-	if strings.TrimSpace(text) != "" {
-		parts = append(parts, openai.TextContentPart(text))
-	}
-	parts = append(parts, openai.TextContentPart(fileID))
-	if len(parts) == 0 {
-		return msg
-	}
-	return openai.UserMessage(parts)
-}
-
-// buildUserMessageWithFiles 为 user 消息追加 file content parts。
-func buildUserMessageWithFiles(msg openai.ChatCompletionMessageParamUnion, fileIDs []string) openai.ChatCompletionMessageParamUnion {
-	text := ""
-	if s, ok := msg.GetContent().AsAny().(*string); ok && s != nil {
-		text = *s
-	}
-
-	parts := make([]openai.ChatCompletionContentPartUnionParam, 0, len(fileIDs)+1)
-	if strings.TrimSpace(text) != "" {
-		parts = append(parts, openai.TextContentPart(text))
-	}
-	for _, id := range fileIDs {
-		id = strings.TrimSpace(id)
-		if id == "" {
-			continue
-		}
-		parts = append(parts, openai.FileContentPart(openai.ChatCompletionContentPartFileFileParam{FileID: param.NewOpt(id)}))
-	}
-	if len(parts) == 0 {
-		return msg
-	}
-	return openai.UserMessage(parts)
-}
-
 // toSDKTools 将内部 ToolDefinition 列表转换为 openai SDK 的 ChatCompletionToolParam 列表。
 func toSDKTools(tools []ToolDefinition) []openai.ChatCompletionToolParam {
 	if len(tools) == 0 {
@@ -397,6 +289,46 @@ func fromSDKToolCalls(sdkCalls []openai.ChatCompletionMessageToolCall) []ToolCal
 	return out
 }
 
+func normalizePromptMessages(messages []PromptMessage) []PromptMessage {
+	out := make([]PromptMessage, 0, len(messages))
+	for _, m := range messages {
+		role := normalizeRole(m.Role)
+		if role == "" {
+			continue
+		}
+		out = append(out, PromptMessage{
+			Role:       role,
+			Content:    m.Content,
+			ToolCalls:  m.ToolCalls,
+			ToolCallID: m.ToolCallID,
+			Name:       m.Name,
+		})
+	}
+	return out
+}
+
+func normalizeRole(role string) string {
+	r := strings.ToLower(strings.TrimSpace(role))
+	if r != "system" && r != "user" && r != "assistant" && r != "tool" {
+		return ""
+	}
+	return r
+}
+
+func promptMessageLengths(messages []PromptMessage) (int, int) {
+	systemLen := 0
+	userLen := 0
+	for _, m := range messages {
+		switch normalizeRole(m.Role) {
+		case "system":
+			systemLen += len(m.Content)
+		case "user":
+			userLen += len(m.Content)
+		}
+	}
+	return systemLen, userLen
+}
+
 func (c *OpenAICompatibleClient) UploadFile(ctx context.Context, file InputFile, purpose string) (string, error) {
 	if strings.TrimSpace(file.FileName) == "" {
 		return "", fmt.Errorf("empty file name")
@@ -460,71 +392,3 @@ func appendIfMissing(items []string, value string) []string {
 	}
 	return append(items, value)
 }
-
-func nonEmptyIDs(ids []string) []string {
-	if len(ids) == 0 {
-		return nil
-	}
-	out := make([]string, 0, len(ids))
-	seen := map[string]struct{}{}
-	for _, id := range ids {
-		id = strings.TrimSpace(id)
-		if id == "" {
-			continue
-		}
-		if _, ok := seen[id]; ok {
-			continue
-		}
-		seen[id] = struct{}{}
-		out = append(out, id)
-	}
-	return out
-}
-
-func normalizePromptMessages(messages []PromptMessage) []PromptMessage {
-	out := make([]PromptMessage, 0, len(messages))
-	for _, m := range messages {
-		role := normalizeRole(m.Role)
-		if role == "" {
-			continue
-		}
-		out = append(out, PromptMessage{
-			Role:       role,
-			Content:    m.Content,
-			ToolCalls:  m.ToolCalls,
-			ToolCallID: m.ToolCallID,
-			Name:       m.Name,
-		})
-	}
-	return out
-}
-
-func normalizeRole(role string) string {
-	r := strings.ToLower(strings.TrimSpace(role))
-	if r != "system" && r != "user" && r != "assistant" && r != "tool" {
-		return ""
-	}
-	return r
-}
-
-func promptMessageLengths(messages []PromptMessage) (int, int) {
-	systemLen := 0
-	userLen := 0
-	for _, m := range messages {
-		switch normalizeRole(m.Role) {
-		case "system":
-			systemLen += len(m.Content)
-		case "user":
-			userLen += len(m.Content)
-		}
-	}
-	return systemLen, userLen
-}
-
-func (c *OpenAICompatibleClient) normalizedFilePromptMode() string {
-	mode := strings.ToLower(strings.TrimSpace(c.filePromptMode))
-	if mode == "system_fileid" || mode == "system_fileid_url" || mode == "system_fileid_uri" {
-		return "system_fileid_uri"
-	}
-	return "user_content_file_parts"
-}