Security Onion MCP (Community Edition)
An MCP server for Security Onion Community Edition that provides 21 read and write tools for SOC operations, including event querying, alert management, detection tuning, PCAP retrieval, and agent enrollment, without requiring a Pro license.
README
CyberHawk Threat Intel
<p align="center"> <img src="https://media.cyberhawkthreatintel.com/general/1771234479938-y9566.png" alt="CyberHawk Threat Intel" width="160"/> </p>
<h1 align="center">Security Onion MCP (Community Edition)</h1> <p align="center"> <strong>By <a href="https://www.cyberhawkthreatintel.com">CyberHawk Threat Intel</a></strong> · Rudra Verma | Senior Cyber Security Architect & Researcher </p>
<p align="center"> <em>A full Tier 2/3 SOC operator MCP for Security Onion — that works on the FREE Community Edition, where the official (Pro-only) MCP cannot.</em> </p>
✅ FULLY WORKING ON SECURITY ONION COMMUNITY EDITION — NO PAID LICENSE REQUIRED
This is a real, live-tested operator built specifically for the FREE Community Edition. Every tool here was proven against a live Security Onion 2.4.190 CE grid — read + write. You do not need a Pro/Enterprise license, and you do not need the Connect API. The official Security Onion MCP requires Pro; this one runs on the free edition the vast majority of people actually use.
<p align="center"> <img src="https://img.shields.io/badge/Security%20Onion-2.4%20CE-16a34a?style=flat-square" alt="Security Onion CE"/> <img src="https://img.shields.io/badge/No%20Pro%20License-required-brightgreen?style=flat-square" alt="No Pro License"/> <img src="https://img.shields.io/badge/Tools-21%20(read%20%2B%20write)-0066cc?style=flat-square" alt="21 tools"/> <img src="https://img.shields.io/badge/MCP-FastMCP%20(Python%203.12%2B)-8B5CF6?style=flat-square" alt="FastMCP"/> <img src="https://img.shields.io/badge/by-CyberHawk%20Threat%20Intel-0066cc?style=flat-square" alt="CyberHawk"/> <img src="https://img.shields.io/badge/License-Apache%202.0-blue?style=flat-square" alt="License"/> </p>
Why This Exists
The official securityonion-mcp depends on the Connect API (Hydra) feature — which is Pro/Enterprise only. On the free Community Edition it has nothing to connect to, so it simply doesn't work.
This MCP works on Community Edition by talking to the grid the way an analyst already can:
- OpenSearch (HTTPS →
:9200) for all read/hunt tools - SSH (
so-*commands + sudo, host-key pinned) for all write/operate tools
No license, no Connect API. And it's a full operator — 21 tools spanning query, hunt, triage, detection authoring & tuning, PCAP, agent enrollment, and gated grid ops — versus the official server's ~3 read-only tools.
Who is this for? SOC analysts and blue teams running Security Onion Community Edition who want an LLM operator that can actually do things on the grid, not just read a Pro one they don't have.
Official vs. This
| Official securityonion-mcp | This (CE) | |
|---|---|---|
| Works on free Community Edition | ❌ (needs Pro Connect API) | ✅ |
| License required | Pro / Enterprise | None |
| Tools | ~3 (read-only) | 21 (read + write) |
| Deploy / tune detections (Suricata, Sigma, YARA) | ❌ | ✅ |
| Elastic Agent enrollment, PCAP export, grid ops | ❌ | ✅ |
| Backend | Connect API | OpenSearch (read) + SSH so-* (write) |
Honest caveat: the official server is vendor-supported and rides SO's internal plumbing. This one is community/DIY and reaches the same surface via OpenSearch + SSH. For a CE user, it's not "similar" — it's the only thing that runs, and it does far more.
Tools (21)
Connectivity & read/hunt (OpenSearch) — ping, grid_status, query_events (OQL: Lucene +
| groupby / | table / | sortby), get_alerts, get_alert_detail, pivot_connection, zeek_logs,
hunt, list_indices, detections_status, agent_list.
Write / operate (SSH so-*, all gated with confirm) — suricata_add_rule, suricata_remove_rule,
suricata_tune, sigma_deploy, yara_deploy, pcap_retrieve, agent_installer, acknowledge_alert,
case_create, grid_command.
Every write tool defaults to a dry-run, validates before applying (Suricata rules go through
so-suricata-testrule), and returns exactly what changed — honest {"ok": false, ...} on failure,
never a fabricated success.
Install
Requires Python 3.12+ and network reach to your SO manager on 9200 (OpenSearch) and 22 (SSH).
git clone <this-repo> securityonion-mcp && cd securityonion-mcp
python -m venv venv && venv\Scripts\activate # Windows (or: source venv/bin/activate)
pip install -r requirements.txt
# credentials live ONLY in an untracked .env.local (never commit real secrets)
copy .env.example .env.local # then fill it in (see below)
Fill .env.local:
| Var | Value |
|---|---|
SO_API_ENDPOINT |
https://YOUR-SO-MANAGER:9200 |
SO_OS_USER / SO_OS_PASSWORD |
so_elastic + its password (sudo cat /opt/so/conf/elasticsearch/curl.config on the manager) |
SO_SSH_HOST / SO_SSH_USER / SO_SSH_PASSWORD |
manager IP + a sudo-capable account |
SO_SSH_HOSTKEY |
the manager's ed25519 fingerprint — pin it (see .env.example for the one-liner) |
SO_API_VERIFY_SSL |
false for a lab self-signed cert, or set SO_CA_CERT and true |
Register with Claude Code
claude mcp add securityonion -s user \
--env-file /abs/path/to/.env.local \
-- /abs/path/to/venv/bin/python -m security_onion_mcp.server
On Windows the command is the venv's python.exe (e.g. ...\venv\Scripts\python.exe -m security_onion_mcp.server), run with the package directory on PYTHONPATH or launched from it.
Usage Examples
"Ping the grid and show me the top firing rules over the last 7 days." →
ping+query_events("tags:alert | groupby rule.name", "-7d", "now")
"What's noisiest, and tune the worst offender for our benign source." →
get_alerts+suricata_tune(sid, "suppress", {"ip": "...", "reason": "..."})
"Write a Suricata rule for this indicator, test it, and deploy it." →
suricata_add_rule(rule, confirm=True)(validate → local.rules → so-rule-update)
"Generate a Windows Elastic Agent installer for enrollment." →
agent_installer("windows", confirm=True)
"Pull the packets between these two hosts." →
pcap_retrieve(src_ip="...", dst_ip="...")
See TESTING.md for the three validation proofs (ping, read query, gated Suricata write)
with live expected output.
Guardrails
- Secrets never touch tracked files or logs — env-only,
.env.localgit-ignored, redaction on logging. - Writes are gated —
confirm=Falsedefault; each returns exactly what changed. - Validate before apply — Suricata rules pass
so-suricata-testrulebefore any deploy. grid_commandis a hard allow-list — never an arbitrary shell; destructive ops needconfirm+ reason.- Honest results — a non-zero exit or non-2xx returns
{"ok": false, ...}; CE-uncertain paths report the real mechanism used (or an honest "SOC-UI" fallback) rather than faking success. - SSH host-key pinned — MITM on the control channel is refused.
Requirements & Compatibility
- Security Onion 2.4.x (built and validated against 2.4.190 CE). OpenSearch reachable on 9200; a sudo-capable SSH account on the manager.
- Enterprise Plus not required. (If you do have Pro, the official MCP is an option too — this one still adds the write/operate surface.)
Disclaimer
For authorized defensive security operations on Security Onion grids you own or are authorized to administer. All actions (querying events, exporting PCAP, deploying detections) are blue-team operations. Authorized security research & administration only. Unauthorized use is illegal.
Connect with CyberHawk Threat Intel
<p align="center"> <a href="https://www.cyberhawkthreatintel.com"> <img src="https://media.cyberhawkthreatintel.com/general/1771234479938-y9566.png" alt="CyberHawk Threat Intel" width="120"/> </a> </p>
<p align="center"> <strong>🦅 Sign up FREE → <a href="https://www.cyberhawkthreatintel.com">cyberhawkthreatintel.com</a></strong> </p>
<p align="center"> <a href="https://youtube.com/@cyberhawkconsultancy">YouTube @cyberhawkconsultancy</a> · <a href="https://youtube.com/@cyberhawkk">YouTube @cyberhawkk</a> · <a href="https://tiktok.com/@cyberhawkthreatintel">TikTok</a> · <a href="https://x.com/cyberhawkintel">X @cyberhawkintel</a> · <a href="https://t.me/cyberhawkthreatintel">Telegram</a> </p>
<p align="center"> <em>Rudra Verma | Senior Cyber Security Architect & Researcher | CyberHawk Threat Intel</em><br/> <em>Authorized security research & penetration testing only. Unauthorized use is illegal.</em> </p>
<p align="center"> #cyberhawkthreatintel #cyberhawkconsultancy #cyberhawkk #cybersecurity #ethicalhacking #pentesting #redteam #threatintel #infosec </p>
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。
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。
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。
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。
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。
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。
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。
mcp-server-qdrant
这个仓库展示了如何为向量搜索引擎 Qdrant 创建一个 MCP (Managed Control Plane) 服务器的示例。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。