Connect coding agents to OffSeq Threat Radar: CVE records with CVSS, CISA KEV, EPSS history, public exploits and ENISA EUVD; SBOM and lockfile scans; patch prioritisation; IOC checks; and your own Radar hosts and decisions.
Radar is a hosted service. This repository holds the client configuration, a Gemini CLI extension and a Claude Code plugin; the server itself runs at https://radar.offseq.com/mcp (Streamable HTTP). It is listed in the official MCP Registry as com.offseq/radar.
Every call needs a Radar API key (tr_…), sent in the X-API-Key header. Sign up at radar.offseq.com and copy the key from Console → Settings. The free plan allows 30 requests a day; paid plans raise the quota and unlock the decision tools. Details: API docs.
As a plugin (reads RADAR_API_KEY from your environment):
export RADAR_API_KEY=tr_your_key
claude plugin marketplace add offseq/radar-mcp
claude plugin install offseq-radar@offseqOr add the server directly:
claude mcp add --transport http radar https://radar.offseq.com/mcp --header "X-API-Key: tr_your_key"gemini extensions install https://github.com/offseq/radar-mcpThe CLI asks for the key and keeps it in the system keychain.
~/.cursor/mcp.json:
{
"mcpServers": {
"radar": {
"url": "https://radar.offseq.com/mcp",
"headers": { "X-API-Key": "${env:RADAR_API_KEY}" }
}
}
}.vscode/mcp.json:
{
"inputs": [
{ "type": "promptString", "id": "radar-key", "description": "Radar API key", "password": true }
],
"servers": {
"radar": {
"type": "http",
"url": "https://radar.offseq.com/mcp",
"headers": { "X-API-Key": "${input:radar-key}" }
}
}
}cline_mcp_settings.json:
{
"mcpServers": {
"radar": {
"type": "streamableHttp",
"url": "https://radar.offseq.com/mcp",
"headers": { "X-API-Key": "tr_your_key" }
}
}
}Any other client that supports Streamable HTTP and custom headers works the same way.
| Tool | What it does |
|---|---|
get_threat |
One CVE or threat record: CVSS, priority and its factors, CISA KEV, EPSS with 30-day history, exploit maturity, ENISA EUVD, OffSeq Mirage sightings, affected versions and remediation |
search_threats |
Search the corpus, e.g. "exploited Fortinet CVEs this month" or "critical CVEs with rising EPSS" |
get_landscape |
What changed in a window: KEV additions, exploitation not in KEV, new public exploits, EPSS risers, new critical CVEs |
prioritize_cves |
Rank 2 to 50 CVEs into a patch order, with the reason for each position |
match_packages |
Check purls, name + ecosystem + version, or CPEs against known vulnerabilities, with the smallest fixing version |
scan_sbom |
Scan a CycloneDX/SPDX SBOM or a lockfile (npm, pnpm, yarn, Cargo, Go, Poetry, uv, Composer, Bundler, pip) with an optional CI-style gate |
check_iocs |
Look up IPs, domains, URLs and hashes in indicators extracted from threat reports |
list_hosts |
Hosts in your Radar inventory, registered by threat-finder |
list_host_findings |
Findings for one host, riskiest first, with exposure, KEV, EPSS and fix versions |
list_changes |
New signals on CVEs you starred or that affect your hosts |
list_decisions |
Your decision ledger per CVE and whether each decision still holds |
record_decision |
Record a decision (patch now, defer, accept, not affected, mitigated); writes to your account |
All tools except record_decision are read-only.
- "Is CVE-2024-3400 exploited, and is there a public exploit?"
- "Scan this package-lock.json and tell me what to upgrade first."
- "What changed in the vulnerability landscape this week?"
The key and your requests go only to radar.offseq.com, operated by SEQ SIA in Latvia. Privacy policy · Terms · support@offseq.com · Security reports: security@offseq.com
The configuration in this repository is MIT licensed. Radar itself is a commercial hosted service.