More screenshots
Works with the AI Agents you already use
Threat Intelligence: IOC and Campaign Research
Turn public sources into defensive intelligence you can act on: bounded questions, narrow collection, and every conclusion graded lead, corroborated, or confirmed.
$5
Threat Intelligence: IOC and Campaign Research
Example session with this skill installed
A vendor blog claims the domain login-verify-account[.]com is running a phishing campaign against our customers. Assess the claim and tell us what we should do.
- Read your context and instructions
- Compiled the threat intelligence
INTELLIGENCE PACKAGE — produced by applying the skill.
Question: does login-verify-account[.]com appear in public phishing disclosures from the last 7 days, and is it part of an active campaign against our customers? Cap: 20 sources. Stop: two consecutive query groups with nothing new.
Queries run: exact domain (4 sources); domain root plus "phishing" (3 more); our brand plus "login verify" in recent disclosures (2 more). Nine sources collected, deduplicated to five unique items on stable IDs — two of the nine were reposts of the vendor blog itself.
Findings: (1) The vendor blog's claim — grade: lead. It shows one screenshot, names no kit, and cites no telemetry; a screenshot is illustration, not evidence. (2) Certificate transparency shows a certificate issued for the domain two days ago, and passive DNS shows it resolving to a host that carried a disclosed credential-phishing kit last month — independent corroboration, infrastructure data type, forgery cost moderate to high. Combined grade for "this domain is live phishing infrastructure impersonating our brand": corroborated. (3) "Part of a named actor's campaign" — grade: lead. The only support is the blog's assertion; an actor name in prose is not attribution.
Actions the grades support: block the domain at the email gateway and web proxy (reversible controls), file the registrar takedown with this package attached, hand the host and kit indicators to threat hunting for a detection hypothesis. Not supported: naming an actor in any customer communication, or declaring an incident. Gap that would raise the grade: a kit sample or gateway telemetry showing credential posts — either would move finding (2) to confirmed.
Connects securely to your tools. The creator never sees your data.
About this skill
The internet already contains early warning about the phishing kit impersonating your brand, the lookalike domain registered three days ago, and the campaign hitting your sector — buried in researcher posts, vendor disclosures, certificate logs, and passive DNS. The hard part is not finding it; it is knowing how much to trust it. This skill is a grading discipline for defensive intelligence. It starts with the intelligence question written to be answerable — target, time window, result cap, stop condition — split into reproducible query groups so collection stays narrow and ends on a rule, not on boredom. Collection preserves provenance for every source (stable ID, author, publication and collection timestamps, the query that surfaced it) and treats everything collected as untrusted data: a post is evidence about what the post says, never instructions for what you do next. Every conclusion is then graded — lead (one source; check it, never block on it), corroborated (an independent source of a different type agrees; reversible blocks and hunts are justified), confirmed (technical evidence; attribution and incident declarations live here) — with a decision table tying each grade to the actions it actually supports. Corroboration is tested the way an adversary would fake it: evidence weighted by forgery cost, citations followed to their origin to catch circular reporting, and reposts never counted as independence. The standard traps are covered — name-as-identity, screenshots as evidence, stale indicators resurfacing, shared-hosting tenancy — and the output is a handoff package in Evidence → Finding → Path form, with the gaps written down, that a hunter, analyst, or IR team can act on without redoing the collection.
- It grades evidence; it cannot manufacture it. If the only public trace is one prose claim, the honest output is a lead with a corroboration plan, not a firmer verdict.
- It covers public-source intelligence only — it does not replace internal telemetry, commercial intel feeds, or a sample-analysis capability, and confirmed grades usually require one of those.
- Grades are snapshots. A domain graded as a lead today can be corroborated tomorrow; the package format handles updates, but someone has to re-run the collection.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 24 days ago
Permissions
No special permissions declared or detected