Enhanced Search

增强的智能搜索能力，优化搜索结果并提供摘要。类似Tavily Web Search的功能，但基于现有web_search工具构建。

Security Scan

The skill's description matches a search/summary helper, but the instructions claim integrations and persistent local access (memory/docs caching, permanent templates, self‑improving logging) and refer to code files that are not present—this mismatch and the implied access to local data and long‑lived caches are concerning.

Purpose & Capability

The stated purpose (enhanced search built on the platform web_search) is coherent with the instructions to call web_search and produce summaries. However the SKILL.md describes additional capabilities (integrating 'docs' and 'memory' sources, self-improving-agent logging, permanent caches) and lists project source files even though the skill package contains no code files—this discrepancy is unexpected and worth verifying.

Instruction Scope

Runtime instructions explicitly reference using conversation context and multiple sources ('web', 'docs', 'memory') and ask users to create a config under ~/.openclaw/workspace/config. While not explicit about reading arbitrary system files, the 'docs' and 'memory' sources and the planned integrations imply the skill may access local documents, agent memory, and record/search user queries—behaviors not declared in registry metadata and potentially privacy‑sensitive.

Install Mechanism

This is an instruction-only skill with no install spec and no downloaded code, which minimizes install-time risk. However the documentation includes deployment steps that assume local files (cp -r enhanced-search ...) and a main.py that do not exist in the published package.

Credentials

The registry declares no required environment variables or credentials, but SKILL.md documents optional env vars and a config file path. More importantly, the skill's described behavior (accessing 'memory' and 'docs', integrating with a 'self-improving-agent') implies access to local data and possibly persistent storage of user queries—access not reflected in declared requirements and not scoped or limited.

Persistence & Privilege

The skill specifies caching (including '摘要模板缓存：永久'—permanent) and mentions logging/search-pattern recording for self‑improvement. Permanent caches or recording of queries increase long‑term data retention and privacy risk. The skill does not request elevated OS privileges nor set always:true, but its design implies persistent local storage of potentially sensitive user queries without clarified retention/consent controls.

What to consider before installing

This skill appears to do what it says (enhanced searches) but has several red flags you should address before installing: 1) The SKILL.md describes source code files and a main.py that are not included—ask the author for the repository or the actual code. 2) The skill plans to read 'docs' and 'memory' and to log/search user queries and keep permanent caches—confirm what local data it will access, how long caches are kept, and whether sensitive queries are stored. 3) If you proceed, run it in a sandbox or with limited permissions, review or request the code, and remove/inspect any config files (~/.openclaw/workspace/config/enhanced-search.json). 4) If you need stronger assurances, request a signed source repo (or packaged release) and explicit explanations of data retention and the exact sources the skill will access.

🔍 Enhanced Search Skill

基于OpenClaw现有web_search工具的增强搜索能力，提供优化结果和智能摘要。

功能特点

🎯 核心功能

1. 智能搜索优化：自动优化搜索查询，提高结果相关性
2. 结果摘要生成：对搜索结果进行摘要，节省阅读时间
3. 多源整合：整合多个搜索结果，提供全面信息
4. 上下文感知：根据对话上下文调整搜索策略

📊 与web_search的区别

功能	web_search	enhanced-search
基础搜索	✅	✅
结果优化	❌	✅
自动摘要	❌	✅
多源整合	❌	✅
上下文感知	❌	✅

使用方法

基本搜索

搜索 [查询内容]

带上下文的搜索

帮我了解 [主题]，我需要 [具体信息]

深度搜索

深度搜索 [复杂查询]，需要详细信息和来源

实现原理

架构设计

用户查询 → 查询优化 → 并行搜索 → 结果整合 → 摘要生成 → 格式化输出

技术栈

• 基础工具：OpenClaw web_search
• 优化算法：查询扩展、关键词提取
• 摘要模型：基于规则的摘要生成
• 整合逻辑：多结果去重和排序

配置选项

环境变量（可选）

# 搜索优化级别
export ENHANCED_SEARCH_OPTIMIZATION=high  # low|medium|high

# 摘要长度
export ENHANCED_SEARCH_SUMMARY_LENGTH=medium  # short|medium|long

# 结果数量
export ENHANCED_SEARCH_RESULT_COUNT=5  # 1-10

配置文件

创建 ~/.openclaw/workspace/config/enhanced-search.json：

{
  "optimization": "high",
  "summary": true,
  "max_results": 5,
  "sources": ["web", "docs", "memory"],
  "cache_ttl": 3600
}

示例

示例1：基础搜索

用户输入：搜索"鸿蒙智行最新动态"

处理流程：

1. 优化查询："鸿蒙智行 最新消息 2026 动态"
2. 执行搜索：调用web_search
3. 生成摘要：提取关键信息
4. 格式化输出：结构化展示结果

示例2：上下文搜索

对话上下文：用户之前问了关于问界M7的问题

用户输入：搜索"智能驾驶技术"

处理流程：

1. 结合上下文：优化为"问界M7 智能驾驶技术 最新进展"
2. 执行搜索
3. 生成针对性摘要
4. 突出与问界M7相关的信息

性能优化

缓存机制

• 搜索结果缓存：1小时
• 查询优化缓存：24小时
• 摘要模板缓存：永久

并行处理

• 多个搜索查询并行执行
• 摘要生成与结果获取并行

错误处理

常见错误及解决方案

1. 网络超时：自动重试，降低优化级别
2. 无结果：扩展查询，尝试相关关键词
3. API限制：使用缓存结果，提示用户稍后重试

降级策略

• 主功能失败时降级到基础web_search
• 摘要失败时返回原始结果
• 优化失败时使用原始查询

扩展能力

插件系统（规划中）

1. 源插件：添加新的搜索源
2. 优化插件：自定义查询优化算法
3. 摘要插件：不同的摘要风格
4. 输出插件：自定义结果格式

集成能力

1. 与self-improving-agent集成：记录搜索模式和优化效果
2. 与知识库集成：优先搜索本地知识
3. 与工作流集成：作为自动化流水线的一部分

开发指南

项目结构

enhanced-search/
├── SKILL.md              # 技能文档（本文件）
├── search_optimizer.py   # 查询优化器
├── result_summarizer.py  # 结果摘要器
├── cache_manager.py      # 缓存管理器
├── config_loader.py      # 配置加载器
└── main.py              # 主入口

核心模块说明

1. 查询优化器 (search_optimizer.py)

class SearchOptimizer:
    def optimize(self, query, context=None):
        """优化搜索查询"""
        # 1. 关键词提取
        # 2. 查询扩展
        # 3. 上下文融合
        # 4. 返回优化后的查询列表

2. 结果摘要器 (result_summarizer.py)

class ResultSummarizer:
    def summarize(self, results, query):
        """生成结果摘要"""
        # 1. 提取关键信息
        # 2. 去重和排序
        # 3. 生成结构化摘要
        # 4. 返回格式化结果

测试用例

单元测试

def test_search_optimization():
    optimizer = SearchOptimizer()
    optimized = optimizer.optimize("华为汽车")
    assert "鸿蒙智行" in optimized  # 查询扩展测试

集成测试

def test_full_search_flow():
    # 完整搜索流程测试
    query = "智能驾驶技术"
    results = enhanced_search(query)
    assert len(results) > 0
    assert "摘要" in results[0]

部署说明

快速部署

# 1. 复制技能目录
cp -r enhanced-search ~/.openclaw/workspace/skills/

# 2. 测试技能
cd ~/.openclaw/workspace/skills/enhanced-search
python main.py --test

# 3. 集成到OpenClaw
# 技能会自动被OpenClaw加载

验证部署

# 验证技能加载
openclaw skills list | grep enhanced-search

# 测试功能
openclaw skills test enhanced-search

维护指南

日常维护

1. 监控搜索质量：定期检查优化效果
2. 更新关键词库：根据趋势更新查询扩展词库
3. 优化缓存策略：根据使用模式调整缓存时间

性能监控

• 搜索响应时间
• 缓存命中率
• 用户满意度（通过self-improving-agent收集）

版本更新

1. 小版本更新：优化算法和bug修复
2. 大版本更新：添加新功能或重构架构

贡献指南

代码贡献

1. Fork项目
2. 创建功能分支
3. 提交Pull Request
4. 通过测试用例

文档贡献

1. 更新示例和文档
2. 添加使用案例
3. 翻译或本地化

许可证

MIT License

联系方式

• 作者：袭人 (Xi Ren)
• 项目：OpenClaw Enhanced Search Skill
• 创建时间：2026-03-12
• 最后更新：2026-03-12

---

备注：此技能为Tavily Web Search的替代方案，专为无法直接安装clawhub技能的环境设计。