search-intelligence-skill

Use search-intelligence-skill to give any AI agent the ability to search the entire internet like an expert OSINT analyst, SEO engineer, and security researcher combined. All searches flow through your SearXNG instance — zero API keys, full privacy, 90+ engines.

The skill generates optimized dork queries, selects intelligent multi-step search strategies, translates operators across engines, routes queries to the best SearXNG engines, scores results by multi-signal relevance, and learns from results to refine searches automatically.

Setup (once)

Install the package
CODEBLOCK0

Start a SearXNG instance (if you don't have one)
CODEBLOCK1

Enable JSON API in SearXNG settings
CODEBLOCK2

Initialize in code
CODEBLOCK3

Common Commands

Natural language search (the main interface)
CODEBLOCK4

Control search depth
CODEBLOCK5

Security scanning — exposed files and panels
CODEBLOCK6

Security scanning — vulnerability research
CODEBLOCK7

OSINT investigation — people
CODEBLOCK8

OSINT investigation — domains and companies
CODEBLOCK9

SEO analysis
CODEBLOCK10

Academic research
CODEBLOCK11

Code and developer search
CODEBLOCK12

File hunting
CODEBLOCK13

News search
CODEBLOCK14

Image and video search
CODEBLOCK15

Social media search
CODEBLOCK16

Direct dork query (no intent parsing)
CODEBLOCK17

Preview queries without executing them
CODEBLOCK18

Build a custom dork from parameters
CODEBLOCK19

Execute a named strategy against a target
CODEBLOCK20

Batch search — multiple queries at once
CODEBLOCK21

Override engine and category selection
CODEBLOCK22

Working with the SearchReport object
CODEBLOCK23

Working with individual SearchResult objects
CODEBLOCK24

AI Agent Integration

Basic tool handler
CODEBLOCK25

With depth control from agent
CODEBLOCK26

Returning structured data to agent
CODEBLOCK27

OpenAI function calling / tool definition
CODEBLOCK28

LangChain tool wrapper
CODEBLOCK29

Context manager for clean resource handling
CODEBLOCK30

Using Individual Components Directly

IntentParser — analyze queries without searching
CODEBLOCK31

DorkGenerator — generate queries without searching
CODEBLOCK32

ResultAnalyzer — score and analyze results
CODEBLOCK33

SearXNGClient — direct API access
CODEBLOCK34

Quick Reference

Search Depths
CODEBLOCK35

Intent Categories (auto-detected)
CODEBLOCK36

Search Strategies (auto-selected by depth + intent)
CODEBLOCK37

Supported SearXNG Engines (90+)
CODEBLOCK38

Dork Operators (auto-translated across engines)
CODEBLOCK39

Dork Template Library

Security dorks available (by subcategory)
CODEBLOCK40

OSINT dorks available (by subcategory)
CODEBLOCK41

SEO dorks available (by subcategory)
CODEBLOCK42

Academic dorks available (by subcategory)
CODEBLOCK43

Code dorks available (by subcategory)
CODEBLOCK44

Advanced Usage

Cross-engine dork translation
CODEBLOCK45

Result scoring details
CODEBLOCK46

Auto-refinement behavior
CODEBLOCK47

Entity extraction capabilities
CODEBLOCK48

Time range detection
CODEBLOCK49

Constraint extraction
CODEBLOCK50

Pagination
CODEBLOCK51

Rate limiting and retries
CODEBLOCK52

Logging for debugging
CODEBLOCK53

API Methods

Method	Purpose	Returns
INLINECODE1	Full intelligent search pipeline	INLINECODE2
INLINECODE3

SearchReport Properties

Property	Type	Description
INLINECODE17	INLINECODE18	Original natural language query
INLINECODE19

SearchResult Properties

Property	Type	Description
INLINECODE41	INLINECODE42	Result title
INLINECODE43

Troubleshooting

SearXNG not reachable
CODEBLOCK54

CODEBLOCK55

No results returned
CODEBLOCK56

Timeout errors
CODEBLOCK57

Rate limiting (429 errors)
CODEBLOCK58

SSL errors (local development only)
CODEBLOCK59

Wrong intent detected
CODEBLOCK60

Memory usage with large result sets
CODEBLOCK61

How It All Works Together

CODEBLOCK62

Notes

Privacy

- All searches route through YOUR SearXNG instance
Zero API keys required for any engine
No data sent to third-party services (except through SearXNG's engine requests)
SearXNG strips tracking parameters and anonymizes requests

Performance tips

- Reuse the SearchSkill instance across searches (connection pooling)
Use depth="quick" for simple lookups, reserve "deep" / "exhaustive" for research
Set auto_refine=False for speed-critical paths
Use skill.suggest_queries() to preview before executing expensive searches
Batch independent queries with INLINECODE65

Accuracy tips

- Include specific entities in your query (domains, emails, CVEs, names)
Use quoted phrases for exact matching: INLINECODE66
Specify time ranges when freshness matters: INLINECODE67
Use depth="deep" or "exhaustive" for comprehensive coverage
Check report.suggestions for refinement ideas
Check report.intent to verify the skill understood your query correctly

Extending the skill

- Add new dork templates in config.py → INLINECODE73
Add new intent signals in config.py → INLINECODE75
Add new engines in config.py → INLINECODE77
Add new operator translations in config.py → INLINECODE79
Add new strategies in config.py → INLINECODE81
Add new subcategory detection in intent.py → INLINECODE83

Confirm before sensitive operations

- Security scanning dorks may trigger alerts on target domains
OSINT queries may involve personal information — use responsibly
Always validate that the target domain/entity is authorized for testing
This tool is for legitimate research, authorized security testing, and SEO analysis

search-intelligence-skill

使用 search-intelligence-skill 可以让任何 AI 代理具备像专业 OSINT 分析师、SEO 工程师和安全研究员一样搜索整个互联网的能力。所有搜索都通过你的 SearXNG 实例进行——无需任何 API 密钥，完全保护隐私，支持 90+ 搜索引擎。

该技能可生成优化的 Dork 查询、选择智能的多步搜索策略、跨搜索引擎转换操作符、将查询路由到最佳的 SearXNG 引擎、根据多信号相关性对结果进行评分，并从结果中学习以自动优化搜索。

设置（一次性）

安装包
bash

从源码安装（推荐）

git clone https://github.com/mouaad-ops/search-intelligence-skill.git
cd search-intelligence-skill
pip install -e .

或直接使用 pip

pip install search-intelligence-skill # 暂不可用

启动 SearXNG 实例（如果没有）
bash

Docker（最快）

docker run -d \
--name searxng \
-p 8888:8080 \
-e SEARXNG_SECRET=your-secret-key \
searxng/searxng:latest

验证是否运行

curl http://localhost:8888/healthz

在 SearXNG 设置中启用 JSON API
yaml

在 searxng/settings.yml 中——确保搜索格式包含 json

search:
formats:
- html
- json

在代码中初始化
python
from searchintelligenceskill import SearchSkill

默认——localhost:8888

skill = SearchSkill()

自定义实例

skill = SearchSkill( searxng_url=http://localhost:8888, timeout=30.0, max_retries=2, rate_limit=0.5, verify_ssl=True, auto_refine=True, maxrefinerounds=1, )

验证连接

if skill.health_check(): print(✓ SearXNG 可达) else: print(✗ 无法访问 SearXNG——请检查 URL 和端口)

常用命令

自然语言搜索（主要接口）
python
from searchintelligenceskill import SearchSkill

skill = SearchSkill(searxng_url=http://localhost:8888)

只需描述你想要的内容——技能会处理一切：

意图检测、Dork 生成、引擎选择、评分

report = skill.search(在 example.com 上查找暴露的 .env 文件)

打印适合 LLM 的格式化输出

print(report.to_context())

访问结构化结果

for r in report.top(5): print(f[{r.relevance:.1f}] {r.title}) print(f {r.url}) print(f {r.snippet[:200]})

控制搜索深度
python
from searchintelligenceskill import Depth

快速——1-2 个查询，单步，快速查找

report = skill.search(什么是 CORS, depth=quick)

标准——3-6 个查询，多引擎，良好的默认值

report = skill.search(python 异步框架对比, depth=standard)

深度——6-12 个查询，多步策略，彻底研究

report = skill.search(对 target.com 进行安全审计, depth=deep)

穷尽——12+ 个查询，完整的 OSINT 链，全面扫描

report = skill.search(对 suspect-domain.com 进行全面侦察, depth=exhaustive)

安全扫描——暴露的文件和面板
python
report = skill.search(
在 example.com 上查找暴露的 .env 文件、管理面板和目录列表,
depth=deep,
)

print(f意图: {report.intent.category.value}/{report.intent.subcategory})

→ 意图: security/exposed_files

print(f策略: {report.strategy.name})

→ 策略: multi_angle

print(f结果数: {len(report.results)})
for r in report.top(10):
print(f [{r.relevance:.1f}] {r.title} — {r.url})

安全扫描——漏洞研究
python

CVE 研究

report = skill.search(CVE-2024-3094 xz 后门利用细节, depth=deep)

特定技术漏洞

report = skill.search( Apache Struts 2024 远程代码执行漏洞, depth=standard, )

暴露的 API 端点

report = skill.search( 在 target.com 上查找暴露的 Swagger API 文档, depth=deep, )

Git 仓库暴露

report = skill.search( 在 example.com 上查找暴露的 .git 目录, depth=deep, )

OSINT 调查——人员
python

按姓名

report = skill.search(
对 John Doe 进行 OSINT 调查——社交媒体、电子邮件、个人资料,
depth=deep,
)

按电子邮件

report = skill.search( 调查 john.doe@example.com——查找所有账户和提及, depth=exhaustive, )

按用户名

report = skill.search( 查找用户名 @johndoe42 的所有账户, depth=deep, )

按电话号码

report = skill.search( 查询电话号码 +1-555-123-4567, depth=standard, )

OSINT 调查——域名和公司
python

域名侦察

report = skill.search(
对 target.com 进行全面域名侦察——子域名、DNS、证书、技术栈,
depth=exhaustive,
)

公司调查

report = skill.search( 调查公司 Acme Corp——员工、备案、数据泄露, depth=deep, )

IP 地址查询

report = skill.search( 调查 IP 192.168.1.1——开放端口、服务、滥用报告, depth=standard, )

SEO 分析
python

网站索引检查

report = skill.search(
对 example.com 进行 SEO 索引分析,
depth=standard,
)

反向链接研究

report = skill.search( 查找指向 example.com 的反向链接, depth=deep, )

竞争对手分析

report = skill.search( 对 example.com 进行 SEO 竞争对手分析——相关网站、排名关键词, depth=deep, )

技术 SEO 审计

report = skill.search( 对 example.com 进行技术 SEO 检查——站点地图、robots.txt、规范链接、hreflang, depth=deep, )

学术研究
python

查找论文

report = skill.search(
2024 年关于 Transformer 架构缩放定律的最新研究论文,
depth=standard,
)

查找数据集

report = skill.search( 下载情感分析基准数据集 CSV, depth=standard, )

查找作者及其作品

report = skill.search( 作者 Yann LeCun 关于深度学习的出版物, depth=deep, )

代码和开发者搜索
python

查找仓库

report = skill.search(
支持 OCR 的 Python PDF 文本提取库,
depth=standard,
)

查找包

report = skill.search( 用于实时 WebSocket 发布/订阅的 npm 包, depth=standard, )

调试错误

report = skill.search( RuntimeError: CUDA out of memory pytorch 解决方案, depth=standard, )

查找文档

report = skill.search( FastAPI 依赖注入文档示例, depth=quick, )

文件搜索
python

查找特定文件类型

report = skill.search(
机器学习速查表 filetype:pdf,
depth=standard,
)

查找数据集

report = skill.search( 2023 年美国人口普查数据下载 CSV, depth=standard, )

查找配置文件

report = skill.search( 微服务 docker-compose 示例 filetype:yaml, depth=standard, )

新闻搜索
python

近期新闻

report = skill.search(
本周关于 AI 监管的最新新闻,
depth=standard,
)

突发新闻

report = skill.search( 今日网络安全突发新闻, depth=quick, )

新闻分析

report = skill.search( 欧盟 AI 法案对初创公司影响的分析, depth=standard, )

图片和视频搜索
python

图片

report = skill.search(
NASA 火星表面高分辨率照片,
depth=standard,
)

视频

report = skill.search( Kubernetes 部署策略视频教程, depth=standard, )

社交媒体搜索
python

Reddit 讨论

report = skill.search(
reddit 讨论关于最佳自托管 Google Photos 替代品,
depth=standard,
)

论坛帖子

report = skill.search( 论坛

search-intelligence-skill智能搜索技能