firewalla-mcp-server
A read-only MCP server that enables Claude to audit Firewalla network security configurations and monitor network activity. It provides tools to inspect devices, rules, alarms, and traffic flows without making any changes to the Firewalla system.
README
firewalla-mcp-server
A read-only Model Context Protocol (MCP) server that lets Claude audit your Firewalla configuration and network security posture via the Firewalla MSP API.
Read-only by design. This server cannot block/unblock devices, create or modify rules, pause services, or make any changes to your Firewalla. It only observes.
What it does
Exposes 8 tools that Claude can use to inspect your Firewalla — devices on your network, active rules, security alarms, network flows, and block/allow target lists.
Example prompts
Security audits
These prompts treat Claude as a network security professional conducting a structured review of your Firewalla configuration. They work best in Claude Desktop or Claude Code where the Firewalla MCP tools are available.
Full network security audit:
You are a senior network security engineer conducting a comprehensive audit of my home network. Using the Firewalla MCP tools, perform the following review and present your findings in a structured report with severity ratings (Critical / High / Medium / Low / Informational):
- Device inventory — Pull the complete device list. Flag any devices with unrecognized MAC vendors, devices not being monitored, or unexpected router-class devices that could indicate a rogue access point.
- Rule audit — Review all block/allow rules. Identify any allow rules that are overly permissive (broad scope, inbound direction, no device restriction). Flag rules with zero hit counts that may be stale.
- Alarm review — Search recent alarms grouped by type and severity. Identify any patterns (repeated alarms from the same device, alarms from unexpected countries, alarms targeting devices that shouldn't have external exposure).
- Target list coverage — Review which block lists are active. Assess whether the current list configuration provides adequate coverage against common threat categories (malware, C2, phishing, cryptomining, newly registered domains).
Conclude with a prioritized list of recommended actions I should take to improve my network security posture.
Firewall rule gap analysis:
Act as a firewall policy analyst. Pull all of my Firewalla rules and the complete device list, then cross-reference them. I need you to identify: (1) devices that have no rules scoped to them at all — are they relying entirely on global rules, and is that intentional? (2) allow rules that grant inbound access — what devices do they target and is the scope appropriately narrow? (3) block rules that have never fired (hit count = 0) — are they stale or is the threat they guard against simply not present? Present your findings as a table for each category with your assessment and recommended action.
Suspicious traffic investigation:
I want to investigate whether any devices on my network are communicating with unexpected external destinations. Search my recent network flows for any traffic to regions outside the US that was NOT blocked by Firewalla. Group the results by device and destination country. For any device that shows unblocked traffic to unusual regions, cross-reference it against my device list to identify what the device is, then check if there are any alarms associated with it. Summarize your findings with a risk assessment for each flagged device.
Quick queries
These are shorter prompts for everyday monitoring and spot checks:
- "List all devices on my network and flag any that have an unknown MAC vendor or aren't being monitored by Firewalla."
- "Show me all allow rules on my Firewalla. Are any of them scoped too broadly?"
- "What are the top alarm types firing on my network right now? Group them by type and give me a count."
- "Check which Firewalla block lists I have active and how many entries each one has. Am I missing any important categories?"
- "Search for any blocked flows in the last 24 hours and group them by destination country. Which countries are showing up the most?"
- "Pull my Firewalla box info — is it online, what firmware version is it running, and how many active alarms are there right now?"
Tools
| Tool | Description |
|---|---|
firewalla_list_boxes |
Discover Firewalla boxes on the MSP account (model, firmware, online status, device/rule/alarm counts) |
firewalla_list_devices |
Inventory all devices on the network (IP, MAC vendor, device type, online status, monitoring flag) |
firewalla_search_flows |
Search network flows with query filters, grouping, and cursor pagination |
firewalla_search_alarms |
Search active security alarms with query filters, grouping, and cursor pagination |
firewalla_get_alarm |
Fetch full detail for a single alarm by box + alarm ID |
firewalla_list_rules |
Audit configured block/allow rules (action, direction, target, scope, hit count) |
firewalla_list_target_lists |
List block/allow target lists (Firewalla-managed and user-defined) |
firewalla_get_target_list |
Fetch metadata for a single target list by ID |
All tools support response_format: "json" | "markdown" and are annotated readOnlyHint: true.
Prerequisites
-
A Firewalla box linked to an MSP account. Even standalone (non-fleet) boxes use the MSP API — it's the only supported public API.
-
An MSP personal access token. Generate one at:
- Log in to your MSP portal at
https://<your-subdomain>.firewalla.net - Go to Account Settings → Personal Access Tokens
- Create a new token and save it somewhere secure
For detailed setup instructions, see Getting Started with the Firewalla MSP API.
- Log in to your MSP portal at
-
Node.js 18+
Install
git clone https://github.com/productengineered/firewalla-mcp.git
cd firewalla-mcp
npm install
npm run build
Configuration
The server reads two environment variables:
| Variable | Description | Example |
|---|---|---|
FIREWALLA_MSP_DOMAIN |
Your MSP subdomain (no https://, no trailing slash) |
yourname.firewalla.net |
FIREWALLA_MSP_TOKEN |
Personal access token from MSP Account Settings | fwtoken_abc123... |
For local development, copy .env.example to .env and fill in your values:
cp .env.example .env
# edit .env with your real values
Usage with Claude Desktop
Add to your claude_desktop_config.json (usually at ~/Library/Application Support/Claude/claude_desktop_config.json on macOS):
{
"mcpServers": {
"firewalla": {
"command": "node",
"args": ["/absolute/path/to/firewalla-mcp/dist/index.js"],
"env": {
"FIREWALLA_MSP_DOMAIN": "yourname.firewalla.net",
"FIREWALLA_MSP_TOKEN": "your-token-here"
}
}
}
}
Note: Claude Desktop launches with a minimal
PATH. Ifnodeisn't found, use the absolute path to your Node.js binary (e.g. the output ofwhich node).
Restart Claude Desktop after editing the config.
Usage with Claude Code
claude mcp add-json --scope user firewalla '{
"type": "stdio",
"command": "node",
"args": ["/absolute/path/to/firewalla-mcp/dist/index.js"],
"env": {
"FIREWALLA_MSP_DOMAIN": "yourname.firewalla.net",
"FIREWALLA_MSP_TOKEN": "your-token-here"
}
}'
Verify with:
claude mcp list
# firewalla: ... - ✓ Connected
New Claude Code sessions will have firewalla_* tools available automatically.
Development
# Source env for local dev
set -a; source .env; set +a
# Run with auto-reload
npm run dev
# Build
npm run build
# Test with MCP Inspector
npx @modelcontextprotocol/inspector --cli node dist/index.js --method tools/list
Firewalla API documentation
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。