diff --git a/yudao-framework/yudao-spring-boot-starter-biz-tenant/src/main/java/cn/iocoder/yudao/framework/tenant/core/redis/TenantRedisCacheManager.java b/yudao-framework/yudao-spring-boot-starter-biz-tenant/src/main/java/cn/iocoder/yudao/framework/tenant/core/redis/TenantRedisCacheManager.java
index aeea4b589c..86577388b5 100644
--- a/yudao-framework/yudao-spring-boot-starter-biz-tenant/src/main/java/cn/iocoder/yudao/framework/tenant/core/redis/TenantRedisCacheManager.java
+++ b/yudao-framework/yudao-spring-boot-starter-biz-tenant/src/main/java/cn/iocoder/yudao/framework/tenant/core/redis/TenantRedisCacheManager.java
@@ -1,6 +1,7 @@
 package cn.iocoder.yudao.framework.tenant.core.redis;
 
 import cn.hutool.core.collection.CollUtil;
+import cn.hutool.core.util.StrUtil;
 import cn.iocoder.yudao.framework.redis.core.TimeoutRedisCacheManager;
 import cn.iocoder.yudao.framework.tenant.core.context.TenantContextHolder;
 import lombok.extern.slf4j.Slf4j;
@@ -21,6 +22,8 @@ import java.util.Set;
 @Slf4j
 public class TenantRedisCacheManager extends TimeoutRedisCacheManager {
 
+    private static final String SPLIT = "#";
+
     private final Set<String> ignoreCaches;
 
     public TenantRedisCacheManager(RedisCacheWriter cacheWriter,
@@ -32,10 +35,11 @@ public class TenantRedisCacheManager extends TimeoutRedisCacheManager {
 
     @Override
     public Cache getCache(String name) {
+        String[] names = StrUtil.splitToArray(name, SPLIT);
         // 如果开启多租户，则 name 拼接租户后缀
         if (!TenantContextHolder.isIgnore()
-            && TenantContextHolder.getTenantId() != null
-            && !CollUtil.contains(ignoreCaches, name)) {
+                && TenantContextHolder.getTenantId() != null
+                && !CollUtil.contains(ignoreCaches, names[0])) {
             name = name + ":" + TenantContextHolder.getTenantId();
         }
 
@@ -43,4 +47,4 @@ public class TenantRedisCacheManager extends TimeoutRedisCacheManager {
         return super.getCache(name);
     }
 
-}
+}
\ No newline at end of file
diff --git a/yudao-module-ai/pom.xml b/yudao-module-ai/pom.xml
index 65017da345..d295f72ff6 100644
--- a/yudao-module-ai/pom.xml
+++ b/yudao-module-ai/pom.xml
@@ -20,7 +20,8 @@
     </description>
     <properties>
         <spring-ai.version>1.1.0</spring-ai.version>
-        <alibaba-ai.version>1.1.0.0-M5</alibaba-ai.version>
+        <!-- https://mvnrepository.com/artifact/com.alibaba.cloud.ai/spring-ai-alibaba -->
+        <alibaba-ai.version>1.1.0.0-RC1</alibaba-ai.version>
         <tinyflow.version>1.2.6</tinyflow.version>
     </properties>
 
@@ -226,6 +227,11 @@
                     <groupId>com.agentsflex</groupId>
                     <artifactId>agents-flex-store-elasticsearch</artifactId>
                 </exclusion>
+                <exclusion>
+                    <!-- 解决 https://t.zsxq.com/pCBZC 问题 -->
+                    <groupId>com.agentsflex</groupId>
+                    <artifactId>agents-flex-search-engine-es</artifactId>
+                </exclusion>
                 <exclusion>
                     <!-- TODO @芋艿：暂时移除 groovy，和 iot 冲突 -->
                     <groupId>org.codehaus.groovy</groupId>
diff --git a/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/controller/admin/knowledge/vo/segment/AiKnowledgeSegmentPageReqVO.java b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/controller/admin/knowledge/vo/segment/AiKnowledgeSegmentPageReqVO.java
index f53d5be076..dd3b90300b 100644
--- a/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/controller/admin/knowledge/vo/segment/AiKnowledgeSegmentPageReqVO.java
+++ b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/controller/admin/knowledge/vo/segment/AiKnowledgeSegmentPageReqVO.java
@@ -11,7 +11,7 @@ import lombok.Data;
 public class AiKnowledgeSegmentPageReqVO extends PageParam {
 
     @Schema(description = "文档编号", example = "1")
-    private Integer documentId;
+    private Long documentId;
 
     @Schema(description = "分段内容关键字", example = "Java 开发")
     private String content;
diff --git a/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/enums/AiDocumentSplitStrategyEnum.java b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/enums/AiDocumentSplitStrategyEnum.java
new file mode 100644
index 0000000000..f0a9cd21d6
--- /dev/null
+++ b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/enums/AiDocumentSplitStrategyEnum.java
@@ -0,0 +1,53 @@
+package cn.iocoder.yudao.module.ai.enums;
+
+import lombok.AllArgsConstructor;
+import lombok.Getter;
+
+/**
+ * AI 知识库文档切片策略枚举
+ *
+ * @author runzhen
+ */
+@AllArgsConstructor
+@Getter
+public enum AiDocumentSplitStrategyEnum {
+
+    /**
+     * 自动识别文档类型并选择最佳切片策略
+     */
+    AUTO("auto", "自动识别"),
+
+    /**
+     * 基于 Token 数量机械切分（默认策略）
+     */
+    TOKEN("token", "Token 切分"),
+
+    /**
+     * 按段落切分（以双换行符为分隔）
+     */
+    PARAGRAPH("paragraph", "段落切分"),
+
+    /**
+     * Markdown QA 格式专用切片器
+     * 识别二级标题作为问题，保持问答对完整性
+     * 长答案智能切分但保留问题作为上下文
+     */
+    MARKDOWN_QA("markdown_qa", "Markdown QA 切分"),
+
+    /**
+     * 语义化切分，保留句子完整性
+     * 在段落和句子边界处切分，避免截断
+     */
+    SEMANTIC("semantic", "语义切分");
+
+    /**
+     * 策略代码
+     */
+    private final String code;
+
+    /**
+     * 策略名称
+     */
+    private final String name;
+
+}
diff --git a/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/AiKnowledgeSegmentServiceImpl.java b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/AiKnowledgeSegmentServiceImpl.java
index 43c7e9cefe..9d64fcce9f 100644
--- a/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/AiKnowledgeSegmentServiceImpl.java
+++ b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/AiKnowledgeSegmentServiceImpl.java
@@ -4,6 +4,7 @@ import cn.hutool.core.collection.CollUtil;
 import cn.hutool.core.collection.ListUtil;
 import cn.hutool.core.util.ObjUtil;
 import cn.hutool.core.util.StrUtil;
+
 import cn.iocoder.yudao.framework.common.enums.CommonStatusEnum;
 import cn.iocoder.yudao.framework.common.pojo.PageResult;
 import cn.iocoder.yudao.framework.common.util.object.BeanUtils;
@@ -15,8 +16,11 @@ import cn.iocoder.yudao.module.ai.dal.dataobject.knowledge.AiKnowledgeDO;
 import cn.iocoder.yudao.module.ai.dal.dataobject.knowledge.AiKnowledgeDocumentDO;
 import cn.iocoder.yudao.module.ai.dal.dataobject.knowledge.AiKnowledgeSegmentDO;
 import cn.iocoder.yudao.module.ai.dal.mysql.knowledge.AiKnowledgeSegmentMapper;
+import cn.iocoder.yudao.module.ai.enums.AiDocumentSplitStrategyEnum;
 import cn.iocoder.yudao.module.ai.service.knowledge.bo.AiKnowledgeSegmentSearchReqBO;
 import cn.iocoder.yudao.module.ai.service.knowledge.bo.AiKnowledgeSegmentSearchRespBO;
+import cn.iocoder.yudao.module.ai.service.knowledge.splitter.MarkdownQaSplitter;
+import cn.iocoder.yudao.module.ai.service.knowledge.splitter.SemanticTextSplitter;
 import cn.iocoder.yudao.module.ai.service.model.AiModelService;
 import com.alibaba.cloud.ai.dashscope.rerank.DashScopeRerankOptions;
 import com.alibaba.cloud.ai.model.RerankModel;
@@ -39,8 +43,7 @@ import java.util.*;
 
 import static cn.iocoder.yudao.framework.common.exception.util.ServiceExceptionUtil.exception;
 import static cn.iocoder.yudao.framework.common.util.collection.CollectionUtils.convertList;
-import static cn.iocoder.yudao.module.ai.enums.ErrorCodeConstants.KNOWLEDGE_SEGMENT_CONTENT_TOO_LONG;
-import static cn.iocoder.yudao.module.ai.enums.ErrorCodeConstants.KNOWLEDGE_SEGMENT_NOT_EXISTS;
+import static cn.iocoder.yudao.module.ai.enums.ErrorCodeConstants.*;
 import static org.springframework.ai.vectorstore.SearchRequest.SIMILARITY_THRESHOLD_ACCEPT_ALL;
 
 /**
@@ -95,8 +98,9 @@ public class AiKnowledgeSegmentServiceImpl implements AiKnowledgeSegmentService
         AiKnowledgeDO knowledgeDO = knowledgeService.validateKnowledgeExists(documentDO.getKnowledgeId());
         VectorStore vectorStore = getVectorStoreById(knowledgeDO);
 
-        // 2. 文档切片
-        List<Document> documentSegments = splitContentByToken(content, documentDO.getSegmentMaxTokens());
+        // 2. 文档切片（使用自动检测策略）
+        List<Document> documentSegments = splitContentByStrategy(content, documentDO.getSegmentMaxTokens(),
+                AiDocumentSplitStrategyEnum.AUTO, documentDO.getUrl());
 
         // 3.1 存储切片
         List<AiKnowledgeSegmentDO> segmentDOs = convertList(documentSegments, segment -> {
@@ -295,8 +299,10 @@ public class AiKnowledgeSegmentServiceImpl implements AiKnowledgeSegmentService
         // 1. 读取 URL 内容
         String content = knowledgeDocumentService.readUrl(url);
 
-        // 2. 文档切片
-        List<Document> documentSegments = splitContentByToken(content, segmentMaxTokens);
+        // 2.1 自动检测文档类型并选择策略
+        AiDocumentSplitStrategyEnum strategy = detectDocumentStrategy(content, url);
+        // 2.2 文档切片
+        List<Document> documentSegments = splitContentByStrategy(content, segmentMaxTokens, strategy, url);
 
         // 3. 转换为段落对象
         return convertList(documentSegments, segment -> {
@@ -333,11 +339,103 @@ public class AiKnowledgeSegmentServiceImpl implements AiKnowledgeSegmentService
         return getVectorStoreById(knowledge);
     }
 
-    private static List<Document> splitContentByToken(String content, Integer segmentMaxTokens) {
-        TextSplitter textSplitter = buildTokenTextSplitter(segmentMaxTokens);
+    /**
+     * 根据策略切分内容
+     *
+     * @param content 文档内容
+     * @param segmentMaxTokens 分段的最大 Token 数
+     * @param strategy 切片策略
+     * @param url 文档 URL（用于自动检测文件类型）
+     * @return 切片后的文档列表
+     */
+    @SuppressWarnings("EnhancedSwitchMigration")
+    private List<Document> splitContentByStrategy(String content, Integer segmentMaxTokens,
+                                                  AiDocumentSplitStrategyEnum strategy, String url) {
+        // 自动检测策略
+        if (strategy == AiDocumentSplitStrategyEnum.AUTO) {
+            strategy = detectDocumentStrategy(content, url);
+            log.info("[splitContentByStrategy][自动检测到文档策略: {}]", strategy.getName());
+        }
+        // 根据策略切分
+        TextSplitter textSplitter;
+        switch (strategy) {
+            case MARKDOWN_QA:
+                textSplitter = new MarkdownQaSplitter(segmentMaxTokens);
+                break;
+            case SEMANTIC:
+                textSplitter = new SemanticTextSplitter(segmentMaxTokens);
+                break;
+            case PARAGRAPH:
+                textSplitter = new SemanticTextSplitter(segmentMaxTokens, 0); // 段落切分，无重叠
+                break;
+            case TOKEN:
+            default:
+                textSplitter = buildTokenTextSplitter(segmentMaxTokens);
+                break;
+        }
+        // 执行切分
         return textSplitter.apply(Collections.singletonList(new Document(content)));
     }
 
+    /**
+     * 自动检测文档类型并选择切片策略
+     *
+     * @param content 文档内容
+     * @param url 文档 URL
+     * @return 推荐的切片策略
+     */
+    private AiDocumentSplitStrategyEnum detectDocumentStrategy(String content, String url) {
+        if (StrUtil.isEmpty(content)) {
+            return AiDocumentSplitStrategyEnum.TOKEN;
+        }
+        // 1. 检测 Markdown QA 格式
+        if (isMarkdownQaFormat(content, url)) {
+            return AiDocumentSplitStrategyEnum.MARKDOWN_QA;
+        }
+        // 2. 检测普通 Markdown 文档
+        if (isMarkdownDocument(url)) {
+            return AiDocumentSplitStrategyEnum.SEMANTIC;
+        }
+        // 3. 默认使用语义切分（比 Token 切分更智能）
+        return AiDocumentSplitStrategyEnum.SEMANTIC;
+    }
+
+    /**
+     * 检测是否为 Markdown QA 格式
+     * 特征：包含多个二级标题（## ）且标题后紧跟答案内容
+     */
+    private boolean isMarkdownQaFormat(String content, String url) {
+        // 文件扩展名判断
+        if (StrUtil.isNotEmpty(url) && !url.toLowerCase().endsWith(".md")) {
+            return false;
+        }
+
+        // 统计二级标题数量
+        long h2Count = content.lines()
+                .filter(line -> line.trim().startsWith("## "))
+                .count();
+
+        // 要求一：至少包含 2 个二级标题才认为是 QA 格式
+        if (h2Count < 2) {
+            return false;
+        }
+
+        // 要求二：检查标题占比（QA 文档标题行数相对较多），如果二级标题占比超过 10%，认为是 QA 格式
+        long totalLines = content.lines().count();
+        double h2Ratio = (double) h2Count / totalLines;
+        return h2Ratio > 0.1;
+    }
+
+    /**
+     * 检测是否为 Markdown 文档
+     */
+    private boolean isMarkdownDocument(String url) {
+        return StrUtil.endWithAnyIgnoreCase(url, ".md", ".markdown");
+    }
+
+    /**
+     * 构建基于 Token 的文本切片器（原有逻辑保留）
+     */
     private static TextSplitter buildTokenTextSplitter(Integer segmentMaxTokens) {
         return TokenTextSplitter.builder()
                 .withChunkSize(segmentMaxTokens)
diff --git a/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/splitter/MarkdownQaSplitter.java b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/splitter/MarkdownQaSplitter.java
new file mode 100644
index 0000000000..1fbf4f2429
--- /dev/null
+++ b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/splitter/MarkdownQaSplitter.java
@@ -0,0 +1,342 @@
+package cn.iocoder.yudao.module.ai.service.knowledge.splitter;
+
+import cn.hutool.core.collection.CollUtil;
+import cn.hutool.core.util.StrUtil;
+import lombok.AllArgsConstructor;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.ai.transformer.splitter.TextSplitter;
+
+import java.util.ArrayList;
+import java.util.Collections;
+import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+/**
+ * Markdown QA 格式专用切片器
+ *
+ * <p>功能特点：
+ * <ul>
+ *   <li>识别二级标题（## ）作为问题标记</li>
+ *   <li>短 QA 对保持完整（不超过 Token 限制）</li>
+ *   <li>长答案智能切分，每个片段保留完整问题作为上下文</li>
+ *   <li>支持自定义 Token 估算器</li>
+ * </ul>
+ *
+ * @author runzhen
+ */
+@Slf4j
+@SuppressWarnings("SizeReplaceableByIsEmpty")
+public class MarkdownQaSplitter extends TextSplitter {
+
+    /**
+     * 二级标题正则：匹配 "## " 开头的行
+     */
+    private static final Pattern H2_PATTERN = Pattern.compile("^##\\s+(.+)$", Pattern.MULTILINE);
+
+    /**
+     * 段落分隔符：双换行
+     */
+    private static final String PARAGRAPH_SEPARATOR = "\n\n";
+
+    /**
+     * 句子分隔符
+     */
+    private static final Pattern SENTENCE_PATTERN = Pattern.compile("[。！？.!?]\\s*");
+
+    /**
+     * 分段的最大 Token 数
+     */
+    private final int chunkSize;
+
+    /**
+     * Token 估算器（简单实现：中文按字符数，英文按单词数的 1.3 倍）
+     */
+    private final TokenEstimator tokenEstimator;
+
+    public MarkdownQaSplitter(int chunkSize) {
+        this.chunkSize = chunkSize;
+        this.tokenEstimator = new SimpleTokenEstimator();
+    }
+
+    @Override
+    protected List<String> splitText(String text) {
+        if (StrUtil.isEmpty(text)) {
+            return Collections.emptyList();
+        }
+
+        // 解析 QA 对
+        List<QaPair> qaPairs = parseQaPairs(text);
+        if (CollUtil.isEmpty(qaPairs)) {
+            // 如果没有识别到 QA 格式，按段落切分
+            return fallbackSplit(text);
+        }
+
+        // 处理每个 QA 对
+        List<String> result = new ArrayList<>();
+        for (QaPair qaPair : qaPairs) {
+            result.addAll(splitQaPair(qaPair));
+        }
+        return result;
+    }
+
+    /**
+     * 解析 Markdown QA 对
+     *
+     * @param content 文本内容
+     * @return QA 对列表
+     */
+    private List<QaPair> parseQaPairs(String content) {
+        // 找到所有二级标题位置
+        List<QaPair> qaPairs = new ArrayList<>();
+        List<Integer> headingPositions = new ArrayList<>();
+        List<String> questions = new ArrayList<>();
+        Matcher matcher = H2_PATTERN.matcher(content);
+        while (matcher.find()) {
+            headingPositions.add(matcher.start());
+            questions.add(matcher.group(1).trim());
+        }
+        if (CollUtil.isEmpty(headingPositions)) {
+            return qaPairs;
+        }
+
+        // 提取每个 QA 对
+        for (int i = 0; i < headingPositions.size(); i++) {
+            int start = headingPositions.get(i);
+            int end = (i + 1 < headingPositions.size())
+                    ? headingPositions.get(i + 1)
+                    : content.length();
+            String qaText = content.substring(start, end).trim();
+            String question = questions.get(i);
+            // 提取答案部分（去掉问题标题）
+            String answer = qaText.substring(qaText.indexOf('\n') + 1).trim();
+            qaPairs.add(new QaPair(question, answer, qaText));
+        }
+        return qaPairs;
+    }
+
+    /**
+     * 切分单个 QA 对
+     *
+     * @param qaPair QA 对
+     * @return 切分后的文本片段列表
+     */
+    private List<String> splitQaPair(QaPair qaPair) {
+        // 如果整个 QA 对不超过限制，保持完整
+        List<String> chunks = new ArrayList<>();
+        String fullQa = qaPair.fullText;
+        int qaTokens = tokenEstimator.estimate(fullQa);
+        if (qaTokens <= chunkSize) {
+            chunks.add(fullQa);
+            return chunks;
+        }
+
+        // 长答案需要切分
+        log.debug("QA 对超过 Token 限制 ({} > {})，开始智能切分: {}", qaTokens, chunkSize, qaPair.question);
+        List<String> answerChunks = splitLongAnswer(qaPair.answer, qaPair.question);
+        for (String answerChunk : answerChunks) {
+            // 每个片段都包含完整问题
+            String chunkText = "## " + qaPair.question + "\n" + answerChunk;
+            chunks.add(chunkText);
+        }
+        return chunks;
+    }
+
+    /**
+     * 切分长答案
+     *
+     * @param answer 答案文本
+     * @param question 问题文本
+     * @return 切分后的答案片段列表
+     */
+    private List<String> splitLongAnswer(String answer, String question) {
+        List<String> chunks = new ArrayList<>();
+        // 预留问题的 Token 空间
+        String questionHeader = "## " + question + "\n";
+        int questionTokens = tokenEstimator.estimate(questionHeader);
+        int availableTokens = chunkSize - questionTokens - 10; // 预留 10 个 Token 的缓冲
+
+        // 先按段落切分
+        String[] paragraphs = answer.split(PARAGRAPH_SEPARATOR);
+        StringBuilder currentChunk = new StringBuilder();
+        int currentTokens = 0;
+        for (String paragraph : paragraphs) {
+            if (StrUtil.isEmpty(paragraph)) {
+                continue;
+            }
+            int paragraphTokens = tokenEstimator.estimate(paragraph);
+            // 如果单个段落就超过限制，需要按句子切分
+            if (paragraphTokens > availableTokens) {
+                // 先保存当前块
+                if (currentChunk.length() > 0) {
+                    chunks.add(currentChunk.toString().trim());
+                    currentChunk = new StringBuilder();
+                    currentTokens = 0;
+                }
+                // 按句子切分长段落
+                chunks.addAll(splitLongParagraph(paragraph, availableTokens));
+                continue;
+            }
+            // 如果加上这个段落会超过限制
+            if (currentTokens + paragraphTokens > availableTokens && currentChunk.length() > 0) {
+                chunks.add(currentChunk.toString().trim());
+                currentChunk = new StringBuilder();
+                currentTokens = 0;
+            }
+            if (currentChunk.length() > 0) {
+                currentChunk.append("\n\n");
+            }
+            // 添加段落
+            currentChunk.append(paragraph);
+            currentTokens += paragraphTokens;
+        }
+
+        // 添加最后一块
+        if (currentChunk.length() > 0) {
+            chunks.add(currentChunk.toString().trim());
+        }
+        return CollUtil.isEmpty(chunks) ? Collections.singletonList(answer) : chunks;
+    }
+
+    /**
+     * 切分长段落（按句子）
+     *
+     * @param paragraph 段落文本
+     * @param availableTokens 可用的 Token 数
+     * @return 切分后的文本片段列表
+     */
+    private List<String> splitLongParagraph(String paragraph, int availableTokens) {
+        // 按句子切分
+        List<String> chunks = new ArrayList<>();
+        String[] sentences = SENTENCE_PATTERN.split(paragraph);
+
+        // 按句子累积切分
+        StringBuilder currentChunk = new StringBuilder();
+        int currentTokens = 0;
+        for (String sentence : sentences) {
+            if (StrUtil.isEmpty(sentence)) {
+                continue;
+            }
+            int sentenceTokens = tokenEstimator.estimate(sentence);
+            // 如果单个句子就超过限制，强制切分
+            if (sentenceTokens > availableTokens) {
+                if (currentChunk.length() > 0) {
+                    chunks.add(currentChunk.toString().trim());
+                    currentChunk = new StringBuilder();
+                    currentTokens = 0;
+                }
+                chunks.add(sentence.trim());
+                continue;
+            }
+            // 如果加上这个句子会超过限制
+            if (currentTokens + sentenceTokens > availableTokens && currentChunk.length() > 0) {
+                chunks.add(currentChunk.toString().trim());
+                currentChunk = new StringBuilder();
+                currentTokens = 0;
+            }
+            // 添加句子
+            currentChunk.append(sentence);
+            currentTokens += sentenceTokens;
+        }
+
+        // 添加最后一块
+        if (currentChunk.length() > 0) {
+            chunks.add(currentChunk.toString().trim());
+        }
+        return chunks.isEmpty() ? Collections.singletonList(paragraph) : chunks;
+    }
+
+    /**
+     * 降级切分策略（当未识别到 QA 格式时）
+     *
+     * @param content 文本内容
+     * @return 切分后的文本片段列表
+     */
+    private List<String> fallbackSplit(String content) {
+        // 按段落切分
+        List<String> chunks = new ArrayList<>();
+        String[] paragraphs = content.split(PARAGRAPH_SEPARATOR);
+
+        // 按段落累积切分
+        StringBuilder currentChunk = new StringBuilder();
+        int currentTokens = 0;
+        for (String paragraph : paragraphs) {
+            if (StrUtil.isEmpty(paragraph)) {
+                continue;
+            }
+            int paragraphTokens = tokenEstimator.estimate(paragraph);
+            // 如果加上这个段落会超过限制
+            if (currentTokens + paragraphTokens > chunkSize && currentChunk.length() > 0) {
+                chunks.add(currentChunk.toString().trim());
+                currentChunk = new StringBuilder();
+                currentTokens = 0;
+            }
+            // 添加段落
+            if (currentChunk.length() > 0) {
+                currentChunk.append("\n\n");
+            }
+            currentChunk.append(paragraph);
+            currentTokens += paragraphTokens;
+        }
+
+        // 添加最后一块
+        if (currentChunk.length() > 0) {
+            chunks.add(currentChunk.toString().trim());
+        }
+        return chunks.isEmpty() ? Collections.singletonList(content) : chunks;
+    }
+
+    /**
+     * QA 对数据结构
+     */
+    @AllArgsConstructor
+    private static class QaPair {
+
+        String question;
+        String answer;
+        String fullText;
+
+    }
+
+    /**
+     * Token 估算器接口
+     */
+    public interface TokenEstimator {
+
+        int estimate(String text);
+
+    }
+
+    /**
+     * 简单的 Token 估算器实现
+     * 中文：1 字符 ≈ 1 Token
+     * 英文：1 单词 ≈ 1.3 Token
+     */
+    private static class SimpleTokenEstimator implements TokenEstimator {
+
+        @Override
+        public int estimate(String text) {
+            if (StrUtil.isEmpty(text)) {
+                return 0;
+            }
+
+            int chineseChars = 0;
+            int englishWords = 0;
+            // 简单统计中英文
+            for (char c : text.toCharArray()) {
+                if (c >= 0x4E00 && c <= 0x9FA5) {
+                    chineseChars++;
+                }
+            }
+            // 英文单词估算
+            String[] words = text.split("\\s+");
+            for (String word : words) {
+                if (word.matches(".*[a-zA-Z].*")) {
+                    englishWords++;
+                }
+            }
+            return chineseChars + (int) (englishWords * 1.3);
+        }
+    }
+
+}
diff --git a/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/splitter/SemanticTextSplitter.java b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/splitter/SemanticTextSplitter.java
new file mode 100644
index 0000000000..4c7112e9ad
--- /dev/null
+++ b/yudao-module-ai/src/main/java/cn/iocoder/yudao/module/ai/service/knowledge/splitter/SemanticTextSplitter.java
@@ -0,0 +1,301 @@
+package cn.iocoder.yudao.module.ai.service.knowledge.splitter;
+
+import cn.hutool.core.util.StrUtil;
+import lombok.extern.slf4j.Slf4j;
+import org.springframework.ai.transformer.splitter.TextSplitter;
+
+import java.util.ArrayList;
+import java.util.Arrays;
+import java.util.Collections;
+import java.util.List;
+import java.util.regex.Matcher;
+import java.util.regex.Pattern;
+
+/**
+ * 语义化文本切片器
+ *
+ * <p>功能特点：
+ * <ul>
+ *   <li>优先在段落边界（双换行）处切分</li>
+ *   <li>其次在句子边界（句号、问号、感叹号）处切分</li>
+ *   <li>避免在句子中间截断，保持语义完整性</li>
+ *   <li>支持中英文标点符号识别</li>
+ * </ul>
+ *
+ * @author runzhen
+ */
+@Slf4j
+public class SemanticTextSplitter extends TextSplitter {
+
+    /**
+     * 分段的最大 Token 数
+     */
+    private final int chunkSize;
+
+    /**
+     * 段落重叠大小（用于保持上下文连贯性）
+     */
+    private final int chunkOverlap;
+
+    /**
+     * 段落分隔符（按优先级排序）
+     */
+    private static final List<String> PARAGRAPH_SEPARATORS = Arrays.asList(
+            "\n\n\n",    // 三个换行
+            "\n\n",      // 双换行
+            "\n"         // 单换行
+    );
+
+    /**
+     * 句子结束标记（中英文标点）
+     */
+    private static final Pattern SENTENCE_END_PATTERN = Pattern.compile(
+            "[。！？.!?]+[\\s\"'）)】\\]]*"
+    );
+
+    /**
+     * Token 估算器
+     */
+    private final MarkdownQaSplitter.TokenEstimator tokenEstimator;
+
+    public SemanticTextSplitter(int chunkSize, int chunkOverlap) {
+        this.chunkSize = chunkSize;
+        this.chunkOverlap = Math.min(chunkOverlap, chunkSize / 2); // 重叠不超过一半
+        this.tokenEstimator = new SimpleTokenEstimator();
+    }
+
+    public SemanticTextSplitter(int chunkSize) {
+        this(chunkSize, 50); // 默认重叠 50 个 Token
+    }
+
+    @Override
+    protected List<String> splitText(String text) {
+        if (StrUtil.isEmpty(text)) {
+            return Collections.emptyList();
+        }
+        return splitTextRecursive(text);
+    }
+
+    /**
+     * 切分文本（递归策略）
+     *
+     * @param text 待切分文本
+     * @return 切分后的文本块列表
+     */
+    private List<String> splitTextRecursive(String text) {
+        List<String> chunks = new ArrayList<>();
+
+        // 如果文本不超过限制，直接返回
+        int textTokens = tokenEstimator.estimate(text);
+        if (textTokens <= chunkSize) {
+            chunks.add(text.trim());
+            return chunks;
+        }
+
+        // 尝试按不同分隔符切分
+        List<String> splits = null;
+        String usedSeparator = null;
+        for (String separator : PARAGRAPH_SEPARATORS) {
+            if (text.contains(separator)) {
+                splits = Arrays.asList(text.split(Pattern.quote(separator)));
+                usedSeparator = separator;
+                break;
+            }
+        }
+
+        // 如果没有找到段落分隔符，按句子切分
+        if (splits == null || splits.size() == 1) {
+            splits = splitBySentences(text);
+            usedSeparator = ""; // 句子切分不需要分隔符
+        }
+
+        // 合并小片段
+        chunks = mergeSplits(splits, usedSeparator);
+        return chunks;
+    }
+
+    /**
+     * 按句子切分
+     *
+     * @param text 待切分文本
+     * @return 句子列表
+     */
+    private List<String> splitBySentences(String text) {
+        // 使用正则表达式匹配句子结束位置
+        List<String> sentences = new ArrayList<>();
+        int lastEnd = 0;
+        Matcher matcher = SENTENCE_END_PATTERN.matcher(text);
+        while (matcher.find()) {
+            String sentence = text.substring(lastEnd, matcher.end()).trim();
+            if (StrUtil.isNotEmpty(sentence)) {
+                sentences.add(sentence);
+            }
+            lastEnd = matcher.end();
+        }
+
+        // 添加剩余部分
+        if (lastEnd < text.length()) {
+            String remaining = text.substring(lastEnd).trim();
+            if (StrUtil.isNotEmpty(remaining)) {
+                sentences.add(remaining);
+            }
+        }
+        return sentences.isEmpty() ? Collections.singletonList(text) : sentences;
+    }
+
+    /**
+     * 合并切分后的小片段
+     *
+     * @param splits 切分后的片段列表
+     * @param separator 片段间的分隔符
+     * @return 合并后的文本块列表
+     */
+    private List<String> mergeSplits(List<String> splits, String separator) {
+        List<String> chunks = new ArrayList<>();
+        List<String> currentChunks = new ArrayList<>();
+        int currentLength = 0;
+
+        for (String split : splits) {
+            if (StrUtil.isEmpty(split)) {
+                continue;
+            }
+            int splitTokens = tokenEstimator.estimate(split);
+            // 如果单个片段就超过限制，进一步递归切分
+            if (splitTokens > chunkSize) {
+                // 先保存当前累积的块
+                if (!currentChunks.isEmpty()) {
+                    String chunkText = String.join(separator, currentChunks);
+                    chunks.add(chunkText.trim());
+                    currentChunks.clear();
+                    currentLength = 0;
+                }
+                // 递归切分大片段
+                if (!separator.isEmpty()) {
+                    // 如果是段落分隔符，尝试按句子切分
+                    chunks.addAll(splitTextRecursive(split));
+                } else {
+                    // 如果已经是句子级别，强制按字符切分
+                    chunks.addAll(forceSplitLongText(split));
+                }
+                continue;
+            }
+            // 计算加上分隔符的 Token 数
+            int separatorTokens = StrUtil.isEmpty(separator) ? 0 : tokenEstimator.estimate(separator);
+            // 如果加上这个片段会超过限制
+            if (!currentChunks.isEmpty() && currentLength + splitTokens + separatorTokens > chunkSize) {
+                // 保存当前块
+                String chunkText = String.join(separator, currentChunks);
+                chunks.add(chunkText.trim());
+
+                // 处理重叠：保留最后几个片段
+                currentChunks = getOverlappingChunks(currentChunks, separator);
+                currentLength = estimateTokens(currentChunks, separator);
+            }
+            // 添加当前片段
+            currentChunks.add(split);
+            currentLength += splitTokens + separatorTokens;
+        }
+
+        // 添加最后一块
+        if (!currentChunks.isEmpty()) {
+            String chunkText = String.join(separator, currentChunks);
+            chunks.add(chunkText.trim());
+        }
+        return chunks;
+    }
+
+    /**
+     * 获取重叠的片段（用于保持上下文）
+     *
+     * @param chunks 当前片段列表
+     * @param separator 片段间的分隔符
+     * @return 重叠的片段列表
+     */
+    private List<String> getOverlappingChunks(List<String> chunks, String separator) {
+        if (chunkOverlap == 0 || chunks.isEmpty()) {
+            return new ArrayList<>();
+        }
+
+        // 从后往前取片段，直到达到重叠大小
+        List<String> overlapping = new ArrayList<>();
+        int tokens = 0;
+        for (int i = chunks.size() - 1; i >= 0; i--) {
+            String chunk = chunks.get(i);
+            int chunkTokens = tokenEstimator.estimate(chunk);
+            if (tokens + chunkTokens > chunkOverlap) {
+                break;
+            }
+            // 添加到重叠列表前端
+            overlapping.add(0, chunk);
+            tokens += chunkTokens + (StrUtil.isEmpty(separator) ? 0 : tokenEstimator.estimate(separator));
+        }
+        return overlapping;
+    }
+
+    /**
+     * 估算片段列表的总 Token 数
+     *
+     * @param chunks 片段列表
+     * @param separator 片段间的分隔符
+     * @return 总 Token 数
+     */
+    private int estimateTokens(List<String> chunks, String separator) {
+        int total = 0;
+        for (int i = 0; i < chunks.size(); i++) {
+            total += tokenEstimator.estimate(chunks.get(i));
+            if (i < chunks.size() - 1 && StrUtil.isNotEmpty(separator)) {
+                total += tokenEstimator.estimate(separator);
+            }
+        }
+        return total;
+    }
+
+    /**
+     * 强制切分长文本（当语义切分失败时）
+     *
+     * @param text 待切分文本
+     * @return 切分后的文本块列表
+     */
+    private List<String> forceSplitLongText(String text) {
+        List<String> chunks = new ArrayList<>();
+        int charsPerChunk = (int) (chunkSize * 0.8); // 保守估计
+        for (int i = 0; i < text.length(); i += charsPerChunk) {
+            int end = Math.min(i + charsPerChunk, text.length());
+            String chunk = text.substring(i, end);
+            chunks.add(chunk.trim());
+        }
+        log.warn("文本过长，已强制按字符切分，可能影响语义完整性");
+        return chunks;
+    }
+
+    /**
+     * 简单的 Token 估算器实现
+     */
+    private static class SimpleTokenEstimator implements MarkdownQaSplitter.TokenEstimator {
+
+        @Override
+        public int estimate(String text) {
+            if (StrUtil.isEmpty(text)) {
+                return 0;
+            }
+
+            int chineseChars = 0;
+            int englishWords = 0;
+            // 简单统计中英文
+            for (char c : text.toCharArray()) {
+                if (c >= 0x4E00 && c <= 0x9FA5) {
+                    chineseChars++;
+                }
+            }
+            // 英文单词估算
+            String[] words = text.split("\\s+");
+            for (String word : words) {
+                if (word.matches(".*[a-zA-Z].*")) {
+                    englishWords++;
+                }
+            }
+            return chineseChars + (int) (englishWords * 1.3);
+        }
+    }
+
+}
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/FileController.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/FileController.java
index 2789242c27..8081aa04a3 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/FileController.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/FileController.java
@@ -13,6 +13,7 @@ import cn.iocoder.yudao.module.infra.service.file.FileService;
 import io.swagger.v3.oas.annotations.Operation;
 import io.swagger.v3.oas.annotations.Parameter;
 import io.swagger.v3.oas.annotations.Parameters;
+import io.swagger.v3.oas.annotations.media.Schema;
 import io.swagger.v3.oas.annotations.tags.Tag;
 import lombok.extern.slf4j.Slf4j;
 import org.springframework.http.HttpStatus;
@@ -44,6 +45,8 @@ public class FileController {
 
     @PostMapping("/upload")
     @Operation(summary = "上传文件", description = "模式一：后端上传文件")
+    @Parameter(name = "file", description = "文件附件", required = true,
+            schema = @Schema(type = "string", format = "binary"))
     public CommonResult<String> uploadFile(@Valid FileUploadReqVO uploadReqVO) throws Exception {
         MultipartFile file = uploadReqVO.getFile();
         byte[] content = IoUtil.readBytes(file.getInputStream());
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/vo/file/FileUploadReqVO.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/vo/file/FileUploadReqVO.java
index 3ffd6f658b..be23401d18 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/vo/file/FileUploadReqVO.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/file/vo/file/FileUploadReqVO.java
@@ -23,7 +23,14 @@ public class FileUploadReqVO {
     @AssertTrue(message = "文件目录不正确")
     @JsonIgnore
     public boolean isDirectoryValid() {
-        return !StrUtil.containsAny(directory, "..", "/", "\\");
+        return isDirectoryValid(directory);
+    }
+
+    public static boolean isDirectoryValid(String directory) {
+        // 1. 不能包含 .. 防止目录穿越
+        // 2. 不能以 / 或 \ 开头，防止上传到根目录
+        return !StrUtil.contains(directory, "..")
+                && !StrUtil.startWithAny(directory, "/", "\\");
     }
 
 }
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/job/vo/log/JobLogRespVO.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/job/vo/log/JobLogRespVO.java
index 3574e98d59..1d33cdf278 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/job/vo/log/JobLogRespVO.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/job/vo/log/JobLogRespVO.java
@@ -49,7 +49,7 @@ public class JobLogRespVO {
 
     @Schema(description = "任务状态，参见 JobLogStatusEnum 枚举", requiredMode = Schema.RequiredMode.REQUIRED, example = "1")
     @ExcelProperty(value = "任务状态", converter = DictConvert.class)
-    @DictFormat(DictTypeConstants.JOB_STATUS)
+    @DictFormat(DictTypeConstants.JOB_LOG_STATUS)
     private Integer status;
 
     @Schema(description = "结果数据", example = "执行成功")
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiAccessLogController.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiAccessLogController.java
index 81b0ee9d84..db86fed4f5 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiAccessLogController.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiAccessLogController.java
@@ -11,11 +11,13 @@ import cn.iocoder.yudao.module.infra.controller.admin.logger.vo.apiaccesslog.Api
 import cn.iocoder.yudao.module.infra.dal.dataobject.logger.ApiAccessLogDO;
 import cn.iocoder.yudao.module.infra.service.logger.ApiAccessLogService;
 import io.swagger.v3.oas.annotations.Operation;
+import io.swagger.v3.oas.annotations.Parameter;
 import io.swagger.v3.oas.annotations.tags.Tag;
 import org.springframework.security.access.prepost.PreAuthorize;
 import org.springframework.validation.annotation.Validated;
 import org.springframework.web.bind.annotation.GetMapping;
 import org.springframework.web.bind.annotation.RequestMapping;
+import org.springframework.web.bind.annotation.RequestParam;
 import org.springframework.web.bind.annotation.RestController;
 
 import javax.annotation.Resource;
@@ -36,6 +38,15 @@ public class ApiAccessLogController {
     @Resource
     private ApiAccessLogService apiAccessLogService;
 
+    @GetMapping("/get")
+    @Operation(summary = "获得 API 访问日志")
+    @Parameter(name = "id", description = "编号", required = true, example = "1024")
+    @PreAuthorize("@ss.hasPermission('infra:api-access-log:query')")
+    public CommonResult<ApiAccessLogRespVO> getApiAccessLog(@RequestParam("id") Long id) {
+        ApiAccessLogDO apiAccessLog = apiAccessLogService.getApiAccessLog(id);
+        return success(BeanUtils.toBean(apiAccessLog, ApiAccessLogRespVO.class));
+    }
+
     @GetMapping("/page")
     @Operation(summary = "获得API 访问日志分页")
     @PreAuthorize("@ss.hasPermission('infra:api-access-log:query')")
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiErrorLogController.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiErrorLogController.java
index 50e22df61c..05945f5c28 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiErrorLogController.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/admin/logger/ApiErrorLogController.java
@@ -50,6 +50,15 @@ public class ApiErrorLogController {
         return success(true);
     }
 
+    @GetMapping("/get")
+    @Operation(summary = "获得 API 错误日志")
+    @Parameter(name = "id", description = "编号", required = true, example = "1024")
+    @PreAuthorize("@ss.hasPermission('infra:api-error-log:query')")
+    public CommonResult<ApiErrorLogRespVO> getApiErrorLog(@RequestParam("id") Long id) {
+        ApiErrorLogDO apiErrorLog = apiErrorLogService.getApiErrorLog(id);
+        return success(BeanUtils.toBean(apiErrorLog, ApiErrorLogRespVO.class));
+    }
+
     @GetMapping("/page")
     @Operation(summary = "获得 API 错误日志分页")
     @PreAuthorize("@ss.hasPermission('infra:api-error-log:query')")
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/AppFileController.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/AppFileController.java
index 7e354fa05b..5805a0fcd0 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/AppFileController.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/AppFileController.java
@@ -9,6 +9,7 @@ import cn.iocoder.yudao.module.infra.service.file.FileService;
 import io.swagger.v3.oas.annotations.Operation;
 import io.swagger.v3.oas.annotations.Parameter;
 import io.swagger.v3.oas.annotations.Parameters;
+import io.swagger.v3.oas.annotations.media.Schema;
 import io.swagger.v3.oas.annotations.tags.Tag;
 import lombok.extern.slf4j.Slf4j;
 import org.springframework.validation.annotation.Validated;
@@ -33,6 +34,8 @@ public class AppFileController {
 
     @PostMapping("/upload")
     @Operation(summary = "上传文件")
+    @Parameter(name = "file", description = "文件附件", required = true,
+            schema = @Schema(type = "string", format = "binary"))
     @PermitAll
     public CommonResult<String> uploadFile(AppFileUploadReqVO uploadReqVO) throws Exception {
         MultipartFile file = uploadReqVO.getFile();
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/vo/AppFileUploadReqVO.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/vo/AppFileUploadReqVO.java
index ec7764e4ca..349b444628 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/vo/AppFileUploadReqVO.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/controller/app/file/vo/AppFileUploadReqVO.java
@@ -1,6 +1,6 @@
 package cn.iocoder.yudao.module.infra.controller.app.file.vo;
 
-import cn.hutool.core.util.StrUtil;
+import cn.iocoder.yudao.module.infra.controller.admin.file.vo.file.FileUploadReqVO;
 import com.fasterxml.jackson.annotation.JsonIgnore;
 import io.swagger.v3.oas.annotations.media.Schema;
 import lombok.Data;
@@ -23,7 +23,7 @@ public class AppFileUploadReqVO {
     @AssertTrue(message = "文件目录不正确")
     @JsonIgnore
     public boolean isDirectoryValid() {
-        return !StrUtil.containsAny(directory, "..", "/", "\\");
+        return FileUploadReqVO.isDirectoryValid(directory);
     }
 
 }
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogService.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogService.java
index 65faa3f333..cc4cda5c52 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogService.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogService.java
@@ -19,6 +19,14 @@ public interface ApiAccessLogService {
      */
     void createApiAccessLog(ApiAccessLogCreateReqDTO createReqDTO);
 
+    /**
+     * 获得 API 访问日志
+     *
+     * @param id 编号
+     * @return API 访问日志
+     */
+    ApiAccessLogDO getApiAccessLog(Long id);
+
     /**
      * 获得 API 访问日志分页
      *
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogServiceImpl.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogServiceImpl.java
index 84ceebdbaa..396068f345 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogServiceImpl.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiAccessLogServiceImpl.java
@@ -45,6 +45,11 @@ public class ApiAccessLogServiceImpl implements ApiAccessLogService {
         }
     }
 
+    @Override
+    public ApiAccessLogDO getApiAccessLog(Long id) {
+        return apiAccessLogMapper.selectById(id);
+    }
+
     @Override
     public PageResult<ApiAccessLogDO> getApiAccessLogPage(ApiAccessLogPageReqVO pageReqVO) {
         return apiAccessLogMapper.selectPage(pageReqVO);
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogService.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogService.java
index b05ccf3d89..5c45797e35 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogService.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogService.java
@@ -19,6 +19,14 @@ public interface ApiErrorLogService {
      */
     void createApiErrorLog(ApiErrorLogCreateReqDTO createReqDTO);
 
+    /**
+     * 获得 API 错误日志
+     *
+     * @param id 编号
+     * @return API 错误日志
+     */
+    ApiErrorLogDO getApiErrorLog(Long id);
+
     /**
      * 获得 API 错误日志分页
      *
diff --git a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogServiceImpl.java b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogServiceImpl.java
index bde4ee54e9..d1bb5ddecd 100644
--- a/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogServiceImpl.java
+++ b/yudao-module-infra/src/main/java/cn/iocoder/yudao/module/infra/service/logger/ApiErrorLogServiceImpl.java
@@ -58,6 +58,11 @@ public class ApiErrorLogServiceImpl implements ApiErrorLogService {
         return apiErrorLogMapper.selectPage(pageReqVO);
     }
 
+    @Override
+    public ApiErrorLogDO getApiErrorLog(Long id) {
+        return apiErrorLogMapper.selectById(id);
+    }
+
     @Override
     public void updateApiErrorLogProcess(Long id, Integer processStatus, Long processUserId) {
         ApiErrorLogDO errorLog = apiErrorLogMapper.selectById(id);
diff --git a/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/sku/ProductSkuMapper.java b/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/sku/ProductSkuMapper.java
index da920a5b82..21f58f75e4 100755
--- a/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/sku/ProductSkuMapper.java
+++ b/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/sku/ProductSkuMapper.java
@@ -5,9 +5,8 @@ import cn.iocoder.yudao.framework.mybatis.core.mapper.BaseMapperX;
 import cn.iocoder.yudao.framework.mybatis.core.query.LambdaQueryWrapperX;
 import cn.iocoder.yudao.module.product.dal.dataobject.sku.ProductSkuDO;
 import com.baomidou.mybatisplus.core.conditions.update.LambdaUpdateWrapper;
-import org.apache.ibatis.annotations.Mapper;
-import org.apache.ibatis.annotations.Param;
-import org.apache.ibatis.annotations.Select;
+import com.baomidou.mybatisplus.extension.handlers.JacksonTypeHandler;
+import org.apache.ibatis.annotations.*;
 
 import java.util.Collection;
 import java.util.List;
@@ -15,7 +14,14 @@ import java.util.List;
 @Mapper
 public interface ProductSkuMapper extends BaseMapperX<ProductSkuDO> {
 
+    /**
+     * 查询商品 SKU（包含已删除）
+     * 注意：使用 @Results 手动指定 typeHandler，否则 @Select 不会应用 autoResultMap，properties 字段无法解析 JSON
+     */
     @Select("SELECT * FROM product_sku WHERE id = #{id}")
+    @Results({
+            @Result(column = "properties", property = "properties", typeHandler = JacksonTypeHandler.class),
+    })
     ProductSkuDO selectByIdIncludeDeleted(@Param("id") Long id);
 
     default List<ProductSkuDO> selectListBySpuId(Long spuId) {
diff --git a/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/spu/ProductSpuMapper.java b/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/spu/ProductSpuMapper.java
index fc00ae78d4..24ebd6864e 100755
--- a/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/spu/ProductSpuMapper.java
+++ b/yudao-module-mall/yudao-module-product/src/main/java/cn/iocoder/yudao/module/product/dal/mysql/spu/ProductSpuMapper.java
@@ -4,15 +4,15 @@ import cn.hutool.core.util.ObjectUtil;
 import cn.iocoder.yudao.framework.common.pojo.PageResult;
 import cn.iocoder.yudao.framework.mybatis.core.mapper.BaseMapperX;
 import cn.iocoder.yudao.framework.mybatis.core.query.LambdaQueryWrapperX;
+import cn.iocoder.yudao.framework.mybatis.core.type.IntegerListTypeHandler;
 import cn.iocoder.yudao.module.product.controller.admin.spu.vo.ProductSpuPageReqVO;
 import cn.iocoder.yudao.module.product.controller.app.spu.vo.AppProductSpuPageReqVO;
 import cn.iocoder.yudao.module.product.dal.dataobject.spu.ProductSpuDO;
 import cn.iocoder.yudao.module.product.enums.ProductConstants;
 import cn.iocoder.yudao.module.product.enums.spu.ProductSpuStatusEnum;
 import com.baomidou.mybatisplus.core.conditions.update.LambdaUpdateWrapper;
-import org.apache.ibatis.annotations.Mapper;
-import org.apache.ibatis.annotations.Param;
-import org.apache.ibatis.annotations.Select;
+import com.baomidou.mybatisplus.extension.handlers.JacksonTypeHandler;
+import org.apache.ibatis.annotations.*;
 
 import java.util.Objects;
 import java.util.Set;
@@ -20,7 +20,15 @@ import java.util.Set;
 @Mapper
 public interface ProductSpuMapper extends BaseMapperX<ProductSpuDO> {
 
+    /**
+     * 查询商品 SPU（包含已删除）
+     * 注意：使用 @Results 手动指定 typeHandler，否则 @Select 不会应用 autoResultMap，sliderPicUrls，deliveryTypes 字段无法解析 JSON
+     */
     @Select("SELECT * FROM product_spu WHERE id = #{id}")
+    @Results({
+            @Result(column = "slider_pic_urls", property = "sliderPicUrls", typeHandler = JacksonTypeHandler.class),
+            @Result(column = "delivery_types", property = "deliveryTypes", typeHandler = IntegerListTypeHandler.class),
+    })
     ProductSpuDO selectByIdIncludeDeleted(@Param("id") Long id);
 
     /**
diff --git a/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/aftersale/core/aop/AfterSaleLogAspect.java b/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/aftersale/core/aop/AfterSaleLogAspect.java
index dc0d538631..a6f1fa5965 100644
--- a/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/aftersale/core/aop/AfterSaleLogAspect.java
+++ b/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/aftersale/core/aop/AfterSaleLogAspect.java
@@ -113,6 +113,9 @@ public class AfterSaleLogAspect {
      * @return 用户类型
      */
     private static Long getUserId() {
+        if (USER_ID.get() != null) {
+            return USER_ID.get();
+        }
         return ObjectUtil.defaultIfNull(WebFrameworkUtils.getLoginUserId(), TradeOrderLogDO.USER_ID_SYSTEM);
     }
 
diff --git a/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/order/core/aop/TradeOrderLogAspect.java b/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/order/core/aop/TradeOrderLogAspect.java
index ccdf91c617..6a22091214 100644
--- a/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/order/core/aop/TradeOrderLogAspect.java
+++ b/yudao-module-mall/yudao-module-trade/src/main/java/cn/iocoder/yudao/module/trade/framework/order/core/aop/TradeOrderLogAspect.java
@@ -109,6 +109,9 @@ public class TradeOrderLogAspect {
      * @return 用户类型
      */
     private static Long getUserId() {
+        if (USER_ID.get() != null) {
+            return USER_ID.get();
+        }
         return ObjectUtil.defaultIfNull(WebFrameworkUtils.getLoginUserId(), TradeOrderLogDO.USER_ID_SYSTEM);
     }
 
diff --git a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/controller/admin/wallet/PayWalletRechargeController.java b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/controller/admin/wallet/PayWalletRechargeController.java
index d2f3afe2fc..339c9bb94f 100644
--- a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/controller/admin/wallet/PayWalletRechargeController.java
+++ b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/controller/admin/wallet/PayWalletRechargeController.java
@@ -51,7 +51,7 @@ public class PayWalletRechargeController {
     public CommonResult<Boolean> updateWalletRechargeRefunded(@RequestBody PayRefundNotifyReqDTO notifyReqDTO) {
         walletRechargeService.updateWalletRechargeRefunded(
                 Long.valueOf(notifyReqDTO.getMerchantOrderId()),
-                Long.valueOf(notifyReqDTO.getMerchantRefundId()),
+                notifyReqDTO.getMerchantRefundId(),
                 notifyReqDTO.getPayRefundId());
         return success(true);
     }
diff --git a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeService.java b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeService.java
index f2ab50ed68..d837b9acd6 100644
--- a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeService.java
+++ b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeService.java
@@ -56,9 +56,9 @@ public interface PayWalletRechargeService {
      * 更新钱包充值记录为已退款
      *
      * @param id          钱包充值记录编号
-     * @param refundId    钱包充值退款编号（实际和 id 相同）
+     * @param refundId    钱包充值退款编号（格式：{id}-refund）
      * @param payRefundId 退款单id
      */
-    void updateWalletRechargeRefunded(Long id, Long refundId, Long payRefundId);
+    void updateWalletRechargeRefunded(Long id, String refundId, Long payRefundId);
 
 }
diff --git a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeServiceImpl.java b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeServiceImpl.java
index 7d9192d876..401ae6e8f2 100644
--- a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeServiceImpl.java
+++ b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletRechargeServiceImpl.java
@@ -220,9 +220,8 @@ public class PayWalletRechargeServiceImpl implements PayWalletRechargeService {
 
     @Override
     @Transactional(rollbackFor = Exception.class)
-    public void updateWalletRechargeRefunded(Long id, Long refundId, Long payRefundId) {
+    public void updateWalletRechargeRefunded(Long id, String refundId, Long payRefundId) {
         // 1.1 获取钱包充值记录
-        // 说明：因为 id 和 refundId 是相同的，所以直接使用 id 查询即可！
         PayWalletRechargeDO walletRecharge = walletRechargeMapper.selectById(id);
         if (walletRecharge == null) {
             log.error("[updateWalletRechargerPaid][钱包充值记录不存在，钱包充值记录 id({})]", id);
@@ -274,8 +273,8 @@ public class PayWalletRechargeServiceImpl implements PayWalletRechargeService {
                     walletRecharge.getId(), payRefundId, toJsonString(walletRecharge), toJsonString(payRefund));
             throw exception(WALLET_RECHARGE_REFUND_FAIL_REFUND_PRICE_NOT_MATCH);
         }
-        // 2.3 校验退款订单商户订单是否匹配
-        if (notEqual(payRefund.getMerchantRefundId(), walletRecharge.getId().toString())) {
+        // 2.3 校验退款订单商户退款单是否匹配
+        if (notEqual(payRefund.getMerchantRefundId(), walletRecharge.getId() + "-refund")) {
             log.error("[validateWalletRechargeCanRefunded][钱包({}) 退款单不匹配({})，请进行处理！payRefund 数据是：{}]",
                     walletRecharge.getId(), payRefundId, toJsonString(payRefund));
             throw exception(WALLET_RECHARGE_REFUND_FAIL_REFUND_ORDER_ID_ERROR);
diff --git a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletServiceImpl.java b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletServiceImpl.java
index 024cecd5ff..776aad13a4 100644
--- a/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletServiceImpl.java
+++ b/yudao-module-pay/src/main/java/cn/iocoder/yudao/module/pay/service/wallet/PayWalletServiceImpl.java
@@ -58,13 +58,22 @@ public class PayWalletServiceImpl implements PayWalletService {
     private PayRefundService refundService;
 
     @Override
+    @SneakyThrows
     public PayWalletDO getOrCreateWallet(Long userId, Integer userType) {
         PayWalletDO wallet = walletMapper.selectByUserIdAndType(userId, userType);
         if (wallet == null) {
-            wallet = new PayWalletDO().setUserId(userId).setUserType(userType)
-                    .setBalance(0).setTotalExpense(0).setTotalRecharge(0);
-            wallet.setCreateTime(LocalDateTime.now());
-            walletMapper.insert(wallet);
+            // 使用双重检查锁，保证钱包创建并发问题
+            // https://gitee.com/zhijiantianya/ruoyi-vue-pro/pulls/1475/files
+            wallet = lockRedisDAO.lock(userId, UPDATE_TIMEOUT_MILLIS, () -> {
+                PayWalletDO newWallet = walletMapper.selectByUserIdAndType(userId, userType);
+                if (newWallet == null) {
+                    newWallet = new PayWalletDO().setUserId(userId).setUserType(userType)
+                            .setBalance(0).setTotalExpense(0).setTotalRecharge(0);
+                    newWallet.setCreateTime(LocalDateTime.now());
+                    walletMapper.insert(newWallet);
+                }
+                return newWallet;
+            });
         }
         return wallet;
     }
diff --git a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/dept/PostController.java b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/dept/PostController.java
index 86815b7a6a..817273bbb4 100644
--- a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/dept/PostController.java
+++ b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/dept/PostController.java
@@ -64,6 +64,14 @@ public class PostController {
         return success(true);
     }
 
+    @DeleteMapping("delete-list")
+    @Operation(summary = "批量删除岗位")
+    @PreAuthorize("@ss.hasPermission('system:post:delete')")
+    public CommonResult<Boolean> deletePostList(@RequestParam("ids") List<Long> ids) {
+        postService.deletePostList(ids);
+        return success(true);
+    }
+
     @GetMapping(value = "/get")
     @Operation(summary = "获得岗位信息")
     @Parameter(name = "id", description = "岗位编号", required = true, example = "1024")
diff --git a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/logger/OperateLogController.java b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/logger/OperateLogController.java
index 0fb7326b76..07ded4c692 100644
--- a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/logger/OperateLogController.java
+++ b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/controller/admin/logger/OperateLogController.java
@@ -13,11 +13,13 @@ import cn.iocoder.yudao.module.system.dal.dataobject.logger.OperateLogDO;
 import cn.iocoder.yudao.module.system.service.logger.OperateLogService;
 import com.fhs.core.trans.anno.TransMethodResult;
 import io.swagger.v3.oas.annotations.Operation;
+import io.swagger.v3.oas.annotations.Parameter;
 import io.swagger.v3.oas.annotations.tags.Tag;
 import org.springframework.security.access.prepost.PreAuthorize;
 import org.springframework.validation.annotation.Validated;
 import org.springframework.web.bind.annotation.GetMapping;
 import org.springframework.web.bind.annotation.RequestMapping;
+import org.springframework.web.bind.annotation.RequestParam;
 import org.springframework.web.bind.annotation.RestController;
 
 import javax.annotation.Resource;
@@ -38,6 +40,15 @@ public class OperateLogController {
     @Resource
     private OperateLogService operateLogService;
 
+    @GetMapping("/get")
+    @Operation(summary = "查看操作日志")
+    @Parameter(name = "id", description = "编号", required = true, example = "1024")
+    @PreAuthorize("@ss.hasPermission('system:operate-log:query')")
+    public CommonResult<OperateLogRespVO> getOperateLog(@RequestParam("id") Long id) {
+        OperateLogDO operateLog = operateLogService.getOperateLog(id);
+        return success(BeanUtils.toBean(operateLog, OperateLogRespVO.class));
+    }
+
     @GetMapping("/page")
     @Operation(summary = "查看操作日志分页列表")
     @PreAuthorize("@ss.hasPermission('system:operate-log:query')")
diff --git a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogService.java b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogService.java
index c647e822d8..bc99a4d3cb 100644
--- a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogService.java
+++ b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogService.java
@@ -20,6 +20,14 @@ public interface OperateLogService {
      */
     void createOperateLog(OperateLogCreateReqDTO createReqDTO);
 
+    /**
+     * 获得操作日志
+     *
+     * @param id 编号
+     * @return 操作日志
+     */
+    OperateLogDO getOperateLog(Long id);
+
     /**
      * 获得操作日志分页列表
      *
diff --git a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogServiceImpl.java b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogServiceImpl.java
index 7e61c6101c..e816fe17aa 100644
--- a/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogServiceImpl.java
+++ b/yudao-module-system/src/main/java/cn/iocoder/yudao/module/system/service/logger/OperateLogServiceImpl.java
@@ -32,6 +32,11 @@ public class OperateLogServiceImpl implements OperateLogService {
         operateLogMapper.insert(log);
     }
 
+    @Override
+    public OperateLogDO getOperateLog(Long id) {
+        return operateLogMapper.selectById(id);
+    }
+
     @Override
     public PageResult<OperateLogDO> getOperateLogPage(OperateLogPageReqVO pageReqVO) {
         return operateLogMapper.selectPage(pageReqVO);