DugganUSA Threat Intelligence

DugganUSA is a threat-intelligence corpus of about 29 million records, built from public feeds such as abuse.ch, Spamhaus, OSV and CISA, from published research credited by name, and from our own GitHub hunting and edge sensors. This connector gives Claude read-only access to it. It writes nothing, needs no account, and sits beside the security tools you already use.

What you can ask:

• Before an install: is this npm, PyPI or RubyGems package, or this GitHub Action, on a malicious-package deny-list or under a published advisory? Answers allow, review or block, with the reason and the record it rests on.
• During triage: what do we know about this IP, domain, URL or file hash? Which actor, which campaign, which feeds, and how confident?
• When patching: where does confirmed exploitation concentrate for this vendor, and is the risk chronic or new?
• For research: search across threat actors, campaigns, CVEs and 1,800+ research posts in plain language.
• For FinOps: what does this model or this agent session cost at list price? Rates are dated and copied from each provider's own pricing page.

Who it is for: AI agent builders who want a pre-install check, security teams triaging alerts, neocloud and GPU-infrastructure operators watching the AI serving stack (inference caches, model servers, theft of metered inference capacity), and anyone who has to put a dollar figure on AI usage.

Absence is not proof of safety: when we hold no record, the tool says so rather than calling it clean. Corrections are published in public at dugganusa.com.