MCP Client B806207c

MCP开发工具

提供方 SRodi

开发工具 未标注 API Key 社区 / 第三方

Netspy是一款集成了AI功能的网络遥测分析工具,提供实时网络监控、进程特定分析和AI驱动的洞察,支持OpenAI函数调用和上下文工具编排。

一句话选型结论

按资料字段整理

MCP Client B806207c 属于开发工具类 MCP 服务器,传输方式未标注,API Key,没有现成安装命令,需按官方文档接入。接入前先确认权限范围;涉及写入、删除或付款的操作,建议先在测试环境验证。

分类开发工具 传输方式未标注 鉴权API Key 部署方式未标注
GitHub 星标 7 仓库热度,无公开仓库时显示未标注
工具数 未标注 暴露给模型的工具数量
资源 / 提示词 0 / 0 资源与提示词模板数量
维护方 社区 / 第三方 核验等级 仅资料核对

工具与能力

开发工具
数量来自来源目录的标注,具体工具清单以官方文档为准
工具数量未标注暴露给模型可调用的工具个数
资源数量0可读取的资源条目
提示词数量0内置提示词模板数量

能力标签

适用场景

网络监控Go本地部署AI分析实时遥测OpenAI集成CLI工具

安装与配置

未标注 API Key
上游未给出传输方式
  1. 1
    准备运行环境

    按官方文档确认运行环境(常见为 Node.js 或 Python),装好后再执行安装命令。

  2. 2
    安装服务

    复制上面的安装命令执行;命令里带占位符的位置,换成自己的密钥或路径。

  3. 3
    写入客户端配置

    在客户端的 MCP 配置里填好启动命令和参数,保存后完全退出客户端再打开,让新配置生效。

  4. 4
    验证是否可用

    连接成功后随便调一个只读工具,确认能返回结果,再接入正式流程。

典型使用场景

3 类
按分类与标签整理的常见用法,具体以官方文档为准
代码与仓库操作

在对话里直接查代码、提 issue、走 PR,省掉来回切工具。

本地命令与脚本

把构建、测试、格式化这类重复命令交给模型执行。

接口联调与测试

让模型调你的接口、跑一遍用例,快速定位问题。

选择建议

选之前先看这几条
按条目字段给出的判断,不替你决定用不用

值得选的地方

  • 基础信息齐全,可以先按官方文档接入试跑。

先想清楚的地方

  • 没有现成安装命令,需要按官方文档自己填启动命令和参数。
  • 使用前要准备好API Key对应的凭据,并按最小权限配置。
  • 资料没标注工具数量,实际能力要装好后确认。
  • 版本更新与行为细节以官方仓库为准。

更适合

  • 需要在开发工具场景里让模型直接调用外部能力的人。
  • 想先快速试一下、能接受在本地或自有环境跑服务的人。

不太适合

  • 不方便管理密钥或第三方授权的人。
  • 需要厂商级 SLA 和技术支持的生产环境。

同类对比

6 个同类
同分类里挑最接近的几条,字段以各自详情页为准
MCP 服务器工具数传输方式部署方式安装命令最后核验
MCP Client B806207c当前 未标注 未标注 未标注 无 2026-05-17
Git 838e2ec1 12 stdio · 本地进程 本地部署 有 2026-05-17
Markitdown MCP 2c653a5f 1 stdio · 本地进程 本地部署 有 2026-05-17
JavaGuide 未标注 stdio · 本地进程 本地部署 有 2026-04-11
awesome-claude-skills 未标注 stdio · 本地进程 本地部署 有 2026-04-26
Context7 831acf40 未标注 stdio · 本地进程 本地部署 有 2026-05-17

服务器档案

开发工具
字段来自项目仓库与公开索引,中文说明由本站标注
提供方SRodi上游标注的维护方 SRodi
分类开发工具本站按上游分类展示
连接方式未标注上游未给出传输方式
鉴权方式API Key远程服务建议用最小权限凭据
维护方社区 / 第三方核验等级 仅资料核对
包名 / 运行时未标注 · 未标注本地安装时用到的包标识与运行时
部署类型未标注上游给出的部署形态
最近核验2026-05-17上游最后核验时间

详细介绍

12.2 KB
来自项目仓库与上游目录的原文,未做删改

Netspy - Contextual Network Telemetry Analyzer

A unified CLI tool that provides MCP (Model Context Protocol) server capabilities for real-time network connectivity analytics with AI-powered insights and OpenAI function calling integration.

🚀 Features

Core Capabilities

  • Real-time Network Monitoring: Track connection attempts, patterns, and packet drops
  • Process-Specific Analysis: Query by PID or process name
  • AI-Powered Insights: Advanced OpenAI function calling with contextual tool orchestration
  • Interactive & Programmatic: Both CLI and API interfaces available
  • Multiple Output Modes: Summary, detailed listings, pattern analysis, and AI insights

AI Integration Highlights

  • OpenAI Function Calling: LLM automatically selects and chains multiple analysis tools
  • Contextual Tool Orchestration: Dynamic tool selection based on query context
  • Comprehensive Analysis: Multi-tool data synthesis for actionable insights
  • Natural Language Queries: Ask questions in plain English about network behavior

🏗️ Architecture```

┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ netspy CLI │───▶│ MCP Server │───▶│ eBPF Server │
│ (MCP Client) │ │ (Internal) │ │ (HTTP API) │
└─────────────────┘ └─────────────────┘ └─────────────────┘
│
▼
┌─────────────────┐
│ OpenAI API │
│ (Function │
│ Calling) │
└─────────────────┘


### Function Calling Flow```
User Query → Conversation Manager → OpenAI API → Function Calls → MCP Tools → Results → Final Response
  1. User Input: Natural language query about network behavior
  2. OpenAI Processing: LLM analyzes query and determines needed tools
  3. Function Calls: System executes appropriate MCP tools with validated parameters
  4. Result Integration: Tool outputs are fed back to the LLM
  5. Final Response: Comprehensive analysis based on real data

📋 Prerequisites

  1. eBPF Network Monitor Server: Build from ebpf-server repositorybash git clone git@github.com:SRodi/ebpf-server.git cd ebpf-server make build # Compiles both Go code AND eBPF programs

  2. Root Privileges: Required for eBPF operations on the server

  3. OpenAI API Key: Required for AI insights (set OPENAI_API_KEY environment variable)bash export OPENAI_API_KEY=your_openai_api_key_here

🛠️ Installation```bash

go build -o netspy ./cmd/netspy


## 🎯 Quick Start```bash
# 1. Start the eBPF API server (run once and keep running)
cd /path/to/ebpf-server
sudo ./bin/ebpf-server --http --port 8080

# 2. Set OpenAI API key
export OPENAI_API_KEY=your_key_here

# 3. Generate some network traffic
curl -s http://google.com

# 4. Use contextual analysis
./netspy
netspy-mcp> contextual "analyze my system"

💬 Usage Examples

Contextual Analysis (Recommended)

The AI-powered analysis automatically selects and chains multiple tools:```bash

Interactive mode with contextual analysis

./netspy
netspy-mcp> contextual "What's happening with my network connections?"
netspy-mcp> contextual "Are there any packet drops or connection issues?"
netspy-mcp> contextual "Analyze the network behavior of process nginx"

Command line mode

./netspy --tool contextual_analysis --query "Analyze my network activity"
./netspy --tool contextual_analysis --query "How is curl behaving?" --process curl


### Traditional Tool Commands```bash
# Interactive mode
./netspy
netspy-mcp> summary --pid 1234 --duration 120
netspy-mcp> list --process curl --max-events 20
netspy-mcp> analyze --process nginx
netspy-mcp> insights "curl made 5 connections in 60 seconds"

# Single command mode
./netspy --tool get_network_summary --pid 1234 --duration 120
./netspy --tool list_connections --process curl --max-events 15
./netspy --tool analyze_patterns --process ssh

🔧 Available Tools

Core Analysis Tools

  • get_network_summary: Aggregated connection statistics for processes
  • list_connections: Recent network connection events with filtering
  • get_packet_drop_summary: Packet loss analysis for connectivity issues
  • list_packet_drops: Detailed packet drop events
  • analyze_patterns: Connection pattern analysis and behavioral insights

AI-Powered Tools

  • contextual_analysis: Advanced AI analysis with automatic tool selection
  • ai_insights: Generate insights from provided summary text

📊 Sample Output

Intelligent Analysis```bash

Contextual Analysis

netspy-mcp> contextual "analyze my system"

Comprehensive Network Analysis

  1. Network Summary:
    - Over the last 60 seconds, there were 10 outbound connection attempts recorded.

  2. Detailed Connection Events:
    - A total of 29 connection events were logged recently.
    - Notable connections include:

    • 10 connections to 127.0.0.1:8080 by the process netspy.
    • Additional connections to DNS servers and local processes.
  3. Packet Drops:
    - There have been 3 packet drops across all processes in the last 60 seconds.

  4. Connection Patterns:
    - Top Destinations: Local connections dominate with 8 connections to :0
    - Protocols Used: Predominantly TCP (9 connections) and UDP (12 connections)

Insights & Recommendations:

  • Monitor netspy: Heavy localhost usage detected
  • Investigate Packet Drops: Monitor for recurring losses
  • Consider Network Capacity: Optimize settings if under heavy load
  • Regular Monitoring: Implement ongoing monitoring for these metrics

### Traditional Commands```bash
netspy-mcp> summary --process curl
Process 'curl' made 5 outbound connection attempts over the last 60 seconds

netspy-mcp> list --max-events 5
Recent connection events (15 total):
  21:05:53 | 127.0.0.1:8080 | TCP | netspy
  21:03:38 | (local socket) | UNIX | snapd
  21:01:08 | (local socket) | UNIX | snapd
  20:58:42 | 192.168.120.2:53 | UDP | systemd-resolve
  20:56:15 | 172.217.164.78:443 | TCP | curl

🎛️ Command Line Options

General Options

  • --server URL: eBPF server URL (default: http://localhost:8080)
  • --verbose: Enable verbose logging
  • --help: Show help information

Tool Execution

  • --tool TOOL: Run specific MCP tool and exit
  • Available tools: get_network_summary, list_connections, get_packet_drop_summary, list_packet_drops, analyze_patterns, ai_insights, contextual_analysis

Tool Parameters

  • --pid PID: Process ID to monitor
  • --process NAME: Process name to monitor
  • --duration SECONDS: Duration in seconds (default: 60)
  • --max-events COUNT: Maximum events to retrieve (default: 100)
  • --summary-text TEXT: Summary text for AI insights
  • --query TEXT: Natural language query for contextual analysis

🤖 AI Function Calling Details

How It Works

The system automatically registers all MCP tools as OpenAI functions, enabling the LLM to:
- Automatically select relevant tools based on user queries
- Chain multiple tools for comprehensive analysis
- Validate parameters and handle errors gracefully
- Maintain conversation context across interactions

Function Parameters

Each tool accepts these parameters (all optional unless specified):

Network Analysis Functions:
- pid (integer): Process ID to analyze
- process_name (string): Process name to analyze
- duration (integer, default: 60): Duration in seconds to analyze
- max_events (integer, default: 10): Maximum number of events to return

AI Functions:
- query (string, required for contextual_analysis): Natural language query
- summary_text (string, required for ai_insights): Summary text to analyze

Key Improvements

  1. No Tool Context Integration → Live Tool Access
    - ✅ LLM has direct access to live network data through function calls

  2. Missing Function Calling → Full OpenAI Function Support
    - ✅ Proper OpenAI function calling with parameter validation
    - ✅ All 7 MCP tools registered as OpenAI functions

  3. Static Approach → Dynamic Tool Usage
    - ✅ LLM contextually selects and chains multiple tools
    - ✅ Demonstrated: 4 tools used automatically for system analysis

  4. Limited Context → Structured Context Management
    - ✅ Multi-turn conversations with full context retention

🔧 Troubleshooting

"Connection refused" or failed connection
- Ensure the eBPF API server is running with --http --port 8080
- Test server health: curl http://localhost:8080/health
- Check that nothing else is using port 8080

"No connections found" when server has data
- Ensure the eBPF API server is running and generating data
- Generate some traffic: curl -s http://google.com
- Use netstat -tlnp | grep 8080 to verify server is listening

"permission denied"
- Use sudo for the eBPF server - eBPF operations require root privileges
- netspy client can run without sudo when using HTTP API mode

OpenAI API errors
- Check that OPENAI_API_KEY environment variable is set
- Verify API key is valid and has sufficient credits
- Check rate limits if experiencing frequent failures

"open bpf/connection.o: no such file or directory"
- Build server with make build in ebpf-server repository
- Server needs compiled eBPF programs, not just Go binary

👨‍💻 Developer Usage

MCP Integration```go

import "github.com/srodi/netspy/internal/mcp"

// Create MCP client with embedded server
mcpClient := mcp.NewMCPClient("http://localhost:8080", true)

// Start interactive mode
ctx := context.Background()
err := mcpClient.StartInteractiveMode(ctx)

// Or execute single command
arguments := map[string]any{
"pid": 1234,
"duration": 120,
}
result, err := mcpClient.RunSingleCommand(ctx, "get_network_summary", arguments)


### OpenAI Integration```go
import "github.com/srodi/netspy/internal/openai"

// Create contextual network analyst
analyst := openai.NewContextualNetworkAnalyst(mcpExecutor)

// Analyze with natural language
analysis, err := analyst.AnalyzeNetworkQuery(ctx, "What's happening with my network?")

// Process-specific analysis
analysis, err := analyst.AnalyzeProcess(ctx, "nginx", 0, 60)

🏗️ Architecture Details

Core Components

  1. Function Call Manager (internal/openai/functions.go)
    - Automatic MCP tool discovery and registration
    - OpenAI function definition generation
    - Parameter validation and type conversion
    - Error handling for function execution

  2. Conversation Manager (internal/openai/client.go)
    - Multi-turn conversation management
    - Automatic function call detection and execution
    - Tool result integration back into conversation
    - Using gpt-4o-mini for optimal function calling

  3. Intelligent Network Analyst (internal/openai/analyst.go)
    - Context-aware query processing
    - Automatic tool selection and chaining
    - Specialized network analysis prompting
    - Conversation history management

  4. MCP Server (internal/mcp/server.go)
    - Model Context Protocol implementation
    - Tool registration and execution
    - HTTP communication with eBPF server

🔮 Future Enhancements

The architecture supports easy extension for:
- Additional MCP tools (automatically discovered)
- Custom analysis workflows
- Different AI model providers
- Streaming responses
- Function call caching
- Multi-modal inputs (images, files)

🚀 Performance

  • Function Call Latency: Sub-second tool execution
  • Tool Coverage: 100% of MCP tools available to LLM
  • Model Efficiency: Using gpt-4o-mini for optimal function calling
  • Zero Code Duplication: Automatic tool discovery eliminates duplicate definitions

📜 License

This project is part of the network telemetry ecosystem and integrates with the ebpf-server for comprehensive network monitoring capabilities.

常见问题

5 问
答案由本站依据条目字段整理,不引用上游原文
MCP Client B806207c 怎么安装?

资料没有给出安装命令,请到官方仓库或文档按说明接入;这类条目多数是生态应用,或尚未发布可安装包。

MCP Client B806207c 怎么在客户端里配置?

它的传输方式未标注,在客户端 MCP 配置里按官方文档填写启动命令和参数,保存后重启客户端生效。

MCP Client B806207c 收费吗?

本站不代收费用,也没有标注具体价格。是否收费取决于服务本身和背后的接口额度,正式使用前按官方说明确认。

MCP Client B806207c 需要 API Key 吗?

资料标注需要API Key认证,使用前准备好对应凭据,并按最小权限配置。

MCP Client B806207c 能提供哪些能力?

资料显示能力标签包括 网络监控、命令行工具、Go、AI分析。具体工具清单以官方文档为准。

热门专题

按标签挑服务器
全部专题 →

服务器信息以作者仓库与官方文档为准;分类、传输方式与鉴权方式的中文说明为中文整理,最近核验 2026-05-17。