Documentation Index
Fetch the complete documentation index at: https://mintlify.com/Mats2208/MCP-Packet-Tracer/llms.txt
Use this file to discover all available pages before exploring further.
MCP Packet Tracer supports two transport modes: stdio, where the AI client spawns the server as a child process on demand (recommended for all desktop clients), and streamable-HTTP, where you start the server yourself and multiple clients or editor windows share the same instance via http://127.0.0.1:39000/mcp. Choose stdio for the simplest zero-configuration experience; choose streamable-HTTP when you need persistent shared state across multiple windows or remote scenarios.
All examples below assume you have already installed the package by running:git clone https://github.com/Mats2208/MCP-Packet-Tracer && cd MCP-Packet-Tracer && pip install -e .
pip install -e ., the packet_tracer_mcp module is importable from any directory, so python -m packet_tracer_mcp --stdio works without cd-ing into the repo first. Choose Your Client
Claude Code
Claude Desktop
VS Code
Cursor
Codex CLI
Generic stdio
Claude Code accepts a single CLI command that registers the server globally in ~/.claude.json, making it available from every directory you launch claude in.claude mcp add --scope user --transport stdio packet-tracer -- python -m packet_tracer_mcp --stdio
--scope user writes to your global profile — no per-project setup needed.- The
-- separator passes everything after it verbatim to the spawned process.
Verify the connection:claude mcp list
# packet-tracer: python -m packet_tracer_mcp --stdio - ✓ Connected
claude mcp remove packet-tracer --scope user
Claude Desktop reads its MCP configuration from a JSON file whose location depends on how Claude Desktop was installed. The Microsoft Store (MSIX) edition sandboxes AppData and uses a different path — a detail the official Anthropic docs do not always mention.Find your config file:| OS / install source | Config path |
|---|
| Windows — installer from claude.ai/download | %APPDATA%\Claude\claude_desktop_config.json |
| Windows — Microsoft Store / MSIX | %LOCALAPPDATA%\Packages\Claude_pzs8sxrjxfjjc\LocalCache\Roaming\Claude\claude_desktop_config.json |
| macOS | ~/Library/Application Support/Claude/claude_desktop_config.json |
| Linux | ~/.config/Claude/claude_desktop_config.json |
WindowsApps\Claude_*, you have MSIX — use the second row in the table above.Get-Process claude | Select-Object -ExpandProperty Path
{
"mcpServers": {
"packet-tracer": {
"command": "python",
"args": ["-m", "packet_tracer_mcp", "--stdio"]
}
}
}
Get-Process claude | Stop-Process -Force
python not found on PATH:The MSIX sandbox may not expose python on PATH. If the MCP indicator never lights up, replace "python" with the absolute path to your interpreter:{
"mcpServers": {
"packet-tracer": {
"command": "C:\\Users\\YOU\\AppData\\Local\\Programs\\Python\\Python312\\python.exe",
"args": ["-m", "packet_tracer_mcp", "--stdio"]
}
}
}
where.exe python (PowerShell) or which python (bash/zsh). Server logs are written to <config-dir>\logs\mcp-server-packet-tracer.log. VS Code (Copilot, Continue, Cline, and other MCP-aware extensions) reads MCP server configuration from .vscode/mcp.json in your workspace root.Option 1 — stdio (recommended): VS Code spawns the server automatically when needed.{
"servers": {
"packet-tracer": {
"type": "stdio",
"command": "python",
"args": ["-m", "packet_tracer_mcp", "--stdio"]
}
}
}
{
"servers": {
"packet-tracer": {
"url": "http://127.0.0.1:39000/mcp"
}
}
}
python -m packet_tracer_mcp
# Serving on http://127.0.0.1:39000/mcp
54321, regardless of which transport mode you choose. Cursor supports both per-user and per-project MCP configuration files.
- Per-user (global):
~/.cursor/mcp.json
- Per-project:
<workspace>/.cursor/mcp.json
{
"mcpServers": {
"packet-tracer": {
"command": "python",
"args": ["-m", "packet_tracer_mcp", "--stdio"]
}
}
}
OpenAI Codex CLI uses TOML, not JSON. Open ~/.codex/config.toml (Windows: %USERPROFILE%\.codex\config.toml) and append the following section:[mcp_servers.packet-tracer]
command = "python"
args = ["-m", "packet_tracer_mcp", "--stdio"]
codex session — Codex picks up MCP servers at launch and will show packet-tracer in the tools list. The vast majority of MCP-compatible clients accept the same three core fields for stdio transport. Translate the values below into whatever syntax your client expects.| Field | Value |
|---|
command | python (or the absolute interpreter path on sandboxed / MSIX systems) |
args | ["-m", "packet_tracer_mcp", "--stdio"] |
transport / type | stdio |
python -m packet_tracer_mcp
http://127.0.0.1:39000/mcp. After Connecting
Once your client shows the server as connected, you can immediately ask your LLM to build a network:
"Create a network with 2 routers, 2 switches, 4 PCs, DHCP and static routing"