Ask your AI assistant what is being exploited right now
Our free MCP server connects Claude, ChatGPT, Cursor and other AI assistants to the live Threat Radar: vulnerabilities confirmed as exploited in the wild, with what to check and do, and a source link for every answer.
From official sources to a sourced answer
- 01Public sourcesOfficial dataCISA KEV and FIRST EPSS
- 02StrikeCyberThreat RadarOur own copy, refreshed hourly
- 03StrikeCyberMCP serverFour read-only tools
- 04Your assistantAI assistantClaude, ChatGPT or Cursor
- 05YouSourced answerLinked to the exact Radar page
What happens when you ask
Your assistant decides when it needs live data, calls one of our tools, and writes the answer from what comes back.
- Step 1
You ask
“Is CVE-2025-61624 being exploited, and what should we do?”
- Step 2
Assistant calls a tool
It sees it needs live data and calls check_cve with CVE-2025-61624.
- Step 3
Server replies
Exploited: yes, with KEV date, EPSS, ransomware use, checks, fixes and the source page.
- Step 4
You get the answer
Plain-English steps to take, with a link to the Radar page to verify.
- Answers come from the live Radar, not the model's training data.
- Every answer links to the page it came from.
- Only the details a tool needs are sent, such as a CVE number.
Three steps. No account, no key.
Copy one address, add it to your assistant, then ask in plain English.
Step 1
Copy the address
No account, password or API key needed.
Server address
https://strikecyber.com/api/mcp/- Freeno sign-up or API key
- 4 toolsall read-only
- Hourlydata refreshed every hour
Example
Is CVE-2025-61624 being exploited?
Yes, confirmed exploited in the wild. Fortinet has a fix: patch now, then check logs for earlier access.
Source: Threat Radar
Works with Claude, ChatGPT, Cursor and VS Code.
Step 2
Add it to your assistant
Pick your assistant, then follow the steps.
Your assistant
- 1Settings, then Connectors
- 2Add custom connector
- 3Paste the address, name it StrikeCyber Threat Radar
- Name
StrikeCyber Threat Radar- Server URL
https://strikecyber.com/api/mcp/
Menus look different? Search settings for “MCP”.
Step 3
Ask in plain English
Click a question to copy it, then paste it in.
Try asking
Every answer links to its Threat Radar page.
Four tools, with questions to try
get_exploited_vulnerabilities Read-onlyLists vulnerabilities confirmed as exploited in the wild, newest first.
Inputs
days- 1 to 365
category- edge devices, email…
vendor- e.g. Fortinet
ransomware_only- true or false
Try asking (click to copy)
check_cve Read-onlySays whether one CVE is exploited, with a plain-English summary, how to check exposure and what to do now.
Inputs
cve- e.g. CVE-2025-61624
Try asking (click to copy)
get_threat_radar_summary Read-onlyA snapshot of the threat picture: counts for 7 and 30 days, ransomware links, top vendors and the technologies under most pressure.
Inputs
None needed. Just ask.
Try asking (click to copy)
explain_security_topic Read-onlyExplains an AI or offensive security topic in plain English from the Field Manual, including how to test for it.
Inputs
topic- e.g. prompt injection, Kerberoasting
Try asking (click to copy)
Read-only by design
It cannot change anything
Every tool only reads public vulnerability data and Field Manual explanations. There are no write actions and no access to your systems.
It cannot see your chats
Your assistant sends only the details a tool needs, such as a CVE number or a vendor name. Your conversation stays with your assistant.
Nothing about you is kept
No accounts and no cookies. Our host keeps standard request logs (IP address and time) for security and the 60-requests-a-minute limit.
Common questions
What is an MCP server?
MCP (Model Context Protocol) is an open standard that lets AI assistants such as Claude, ChatGPT and Cursor call outside tools and data. Adding our MCP server gives your assistant live, sourced answers about exploited vulnerabilities instead of relying on what it learned in training.
Is it free and do I need an account?
It is free, with no account, sign-up or API key. Add the address to your assistant and start asking.
Is it safe to connect?
Yes. Every tool is read-only: it can only return public vulnerability data and Field Manual explanations. It cannot see your files, your chats or your systems, and it cannot change anything.
What do you collect when I use it?
Nothing about your conversation. The server only receives the tool call your assistant makes, such as a CVE number, and our host keeps standard request logs (IP address and time) for security and rate limiting.
Where does the data come from?
The CISA Known Exploited Vulnerabilities catalogue, FIRST EPSS exploitation scores, refreshed hourly, plus guidance written by our operators.
Are there limits?
Up to 60 requests a minute from one address, which is far more than a person chatting needs. Lists return up to 50 vulnerabilities per call.
Other ways to use the data
Not sure which of these you expose?
A free Cyber Security Snapshot maps your internet-facing systems, so you know which exploited vulnerabilities apply to you.