NetMCP
Enables AI-powered optimization of Linux network performance through natural language commands, with tools for discovery, planning, validation, and safe execution.
README
<div align="center">
🌐 NetMCP
AI-Powered Linux Network Optimization via Model Context Protocol
Transform natural language into optimized network configurations
🔴 Live Server · Getting Started · Features · Profiles · Architecture · Tools
</div>
🔴 Live Server
NetMCP is live in production on FastMCP Cloud! Connect your MCP client to:
https://netmcp.fastmcp.app/mcp
No installation required—just point your AI assistant to the hosted server and start optimizing.
🎯 Overview
NetMCP is an MCP server that enables AI assistants like Claude to intelligently optimize Linux network performance. It bridges the gap between high-level optimization goals ("optimize my network for gaming") and low-level Linux commands (sysctl, tc, nft).
┌──────────────────┐ ┌──────────────────┐ ┌──────────────────┐
│ AI Assistant │ MCP │ NetMCP │ │ Linux Kernel │
│ (Claude, etc) │◄────►│ Server │─────►│ Network Stack │
└──────────────────┘ └──────────────────┘ └──────────────────┘
▲ │
│ ▼
Natural Language 29 Config Cards
"Reduce my latency" 5 Profiles
40+ Tools
✨ Why NetMCP?
| Traditional Approach | With NetMCP |
|---|---|
Manual sysctl tuning |
Declarative optimization plans |
| Copy-paste commands from forums | Research-backed profiles |
| No rollback on failure | Automatic checkpoints & rollback |
| Trial and error | Validated against safety policies |
| Root access chaos | Controlled privileged execution |
🚀 Getting Started
Prerequisites
- Linux (Ubuntu 20.04+, Debian 11+, or similar)
- Python 3.10+
- Network tools:
ip,sysctl,tc,nft
Installation
git clone https://github.com/SurriyaGokul/mcp-net-optimizer.git
cd mcp-net-optimizer
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
# Enable privileged commands (one-time setup)
./setup_sudo.sh
Configure with Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"netmcp": {
"command": "python",
"args": ["-m", "server.main"],
"cwd": "/path/to/mcp-net-optimizer"
}
}
}
Run Standalone
python -m server.main
⚡ Features
<table> <tr> <td width="50%">
🎴 29 Configuration Cards
Full coverage of the Linux networking stack:
- 16 Sysctl cards — TCP tuning, buffers, congestion control
- 8 Traffic Control cards — QoS, shaping, queuing
- 5 Firewall cards — Rate limiting, connection tracking, NAT
</td> <td width="50%">
🛡️ Enterprise Safety
Production-ready security features:
- Command allowlisting — Only approved binaries execute
- Automatic checkpoints — Snapshot before every change
- Instant rollback — One command to restore state
- Audit logging — Complete execution history
</td> </tr> <tr> <td>
🔧 40+ MCP Tools
Comprehensive network management:
- Discovery — Interfaces, routes, DNS, latency tests
- Planning — Type-safe optimization plans
- Validation — Policy enforcement before execution
- Execution — Atomic apply with rollback
- Benchmarking — Before/after performance comparison
</td> <td>
🔐 Sudo Management
Flexible privilege escalation:
- Passwordless setup — Run
./setup_sudo.shonce - Cached credentials — Authenticate once, cached 15 min
- MCP tools —
check_sudo_access,request_sudo_access - Secure by default — Only allowlisted commands
</td> </tr> </table>
🎮 Optimization Profiles
Five research-backed profiles optimized for specific workloads:
| Profile | Focus | Key Optimizations | Target Metrics |
|---|---|---|---|
| 🎮 Gaming | Ultra-low latency | BBR, tcp_low_latency, fq qdisc | <20ms p95, <5ms jitter |
| 📺 Streaming | Max throughput | Large buffers, BBR, HTB shaping | 90%+ link utilization |
| 📞 Video Calls | Balanced | DSCP marking, moderate buffers | <150ms latency (ITU-T) |
| 📦 Bulk Transfer | Maximum bandwidth | 128MB buffers, aggressive BBR | >1Gbps sustained |
| 🖥️ Server | High concurrency | SYN cookies, conntrack, rate limits | 10K+ connections |
Example: Gaming Optimization
plan = {
"iface": "eth0",
"profile": "gaming",
"changes": {
"sysctl": {
"net.ipv4.tcp_congestion_control": "bbr",
"net.ipv4.tcp_low_latency": "1",
"net.core.default_qdisc": "fq"
},
"qdisc": {"type": "fq"}
}
}
🏗️ Architecture
Pipeline Flow
DISCOVER PLAN VALIDATE RENDER APPLY
┌─────────┐ ┌─────────┐ ┌─────────┐ ┌─────────┐ ┌─────────┐
│ Inspect │───►│ Declare │─────►│ Check │────►│Generate │────►│ Execute │
│ System │ │ Intent │ │ Policies│ │Commands │ │ Safely │
└─────────┘ └─────────┘ └─────────┘ └─────────┘ └─────────┘
│ │ │ │ │
30+ tools Pydantic Policy YAML sysctl/tc/ Checkpoint
No side schemas validation nft scripts + Rollback
effects
Project Structure
mcp-net-optimizer/
├── server/
│ ├── main.py # MCP server entry point
│ ├── registry.py # Tool & resource registration
│ ├── schema/models.py # Pydantic data models
│ └── tools/
│ ├── discovery.py # 30+ system inspection tools
│ ├── planner.py # Plan → Commands renderer
│ ├── validator.py # Policy validation
│ ├── validation_engine.py # Before/after comparison
│ ├── validation_metrics.py # Network benchmarks
│ ├── audit_log.py # Execution logging
│ ├── apply/ # Command executors
│ │ ├── apply.py # Orchestration + rollback
│ │ ├── checkpoints.py
│ │ ├── sysctl.py
│ │ ├── tc.py
│ │ └── nft.py
│ └── util/
│ ├── shell.py # Safe command execution
│ └── policy_loader.py
├── policy/
│ ├── config_cards/ # 29 YAML card definitions
│ ├── profiles.yaml # 5 optimization profiles
│ └── validation_limits.yaml
└── setup_sudo.sh # Passwordless sudo setup
🔧 MCP Tools
Discovery (No Side Effects)
| Tool | Description |
|---|---|
ip_info |
Network interfaces and addresses |
ip_route |
Routing table |
ping_host |
ICMP latency test |
traceroute |
Network path analysis |
tc_qdisc_show |
Traffic control status |
nft_list_ruleset |
Firewall rules |
ss_summary |
Socket statistics |
Planning & Validation
| Tool | Description |
|---|---|
validate_change_plan_tool |
Validate plan against policies |
render_change_plan_tool |
Convert plan to executable commands |
test_network_performance_tool |
Run comprehensive benchmarks |
validate_configuration_changes_tool |
Compare before/after results |
Execution & Safety
| Tool | Description |
|---|---|
apply_rendered_plan_tool |
Execute with checkpoint + rollback |
snapshot_checkpoint_tool |
Manual checkpoint creation |
rollback_to_checkpoint_tool |
Restore previous state |
list_checkpoints_tool |
View available checkpoints |
Privilege Management
| Tool | Description |
|---|---|
check_sudo_access_tool |
Check if sudo is available |
request_sudo_access_tool |
Authenticate for temporary access |
get_sudo_setup_instructions_tool |
Setup help |
📊 Benchmarking & Validation
NetMCP includes a complete validation pipeline to measure optimization impact:
# 1. Run baseline benchmark
before = test_network_performance_tool(profile="gaming")
# 2. Apply optimizations
apply_rendered_plan_tool(rendered_plan)
# 3. Run post-optimization benchmark
after = test_network_performance_tool(profile="gaming")
# 4. Compare and validate
result = validate_configuration_changes_tool(before, after, "gaming")
# → {"decision": "KEEP", "score": 75, "summary": "Latency improved 15%"}
Validation Decisions
| Score | Decision | Action |
|---|---|---|
| ≥60 | KEEP | Changes improved performance |
| 20-59 | UNCERTAIN | Mixed results, review recommended |
| <20 | ROLLBACK | Performance degraded, auto-rollback available |
🔒 Security Model
Command Allowlisting
Only explicitly approved binaries can execute:
# server/config/allowlist.yaml
binaries:
- /usr/sbin/sysctl
- /usr/sbin/tc
- /usr/sbin/nft
- /usr/bin/ping
- /usr/bin/iperf3
# ... etc
Privilege Escalation
Three options for sudo access:
- Permanent (Recommended): Run
./setup_sudo.sh— configures passwordless sudo for network commands only - Session-based: Use
request_sudo_access_tool(password="...")— caches for 15 minutes - Manual: Run
sudo -vin terminal before using MCP
📈 Performance Results
Real-world improvements measured across profiles:
| Profile | Metric | Improvement |
|---|---|---|
| Gaming | Latency | 10-30% reduction |
| Gaming | Jitter | 20-40% reduction |
| Streaming | Throughput | 2-4× on congested links |
| Video Calls | Connection time | 15-25% faster |
| Server | Connection capacity | 10× increase |
🤝 Contributing
Contributions are welcome! Areas of interest:
- Additional configuration cards
- New optimization profiles
- Cross-platform support
- Performance benchmarks
- Documentation improvements
📄 License
MIT License — see LICENSE for details.
<div align="center">
Built with ❤️ for the AI-native networking future
</div>
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.