CyberKit

Recon Pivots

Queries that walk outward from one domain, IP, certificate, or hash on the platforms responders actually pivot through.

One indicator is rarely the whole picture. A domain leads to a certificate, the certificate leads to other hosts, and those hosts lead to the rest of the campaign. These are the queries I run to walk outward until the edges stop moving.

Each row names its platform because the syntax does not travel. Shodan syntax pasted into Censys or a web search returns confident-looking garbage rather than an error.

Before you read the table

  • Swap every <placeholder> for your value before you run the query. The angle brackets are not part of the syntax.
  • Access varies. crt.sh, urlscan.io, the Wayback CDX API, RDAP and abuse.ch search are free with no account. Shodan, Censys and GitHub code search want a sign-in. VirusTotal's search modifiers are a paid tier.
  • Running a query against a third-party platform tells that platform what you searched for. Mind that on a sensitive case.

59 rows.

Replace each <placeholder> before you run a query. Every platform's field names were checked against its own documentation.

Recon pivot queries
GoalPlatformQueryWhy it worksTrack
Hosts serving a certificate for a domainShodanssl.cert.subject.cn:"<domain>"Every scanned host presenting a TLS certificate whose subject common name is this domain, wherever it happens to live. This is how the origin behind a CDN, a staging copy on a different provider, or a box still serving a certificate you thought was retired all surface.Infrastructure pivoting
Certificate data mentioning an organizationShodanssl:"<organization name>"Broader than the common-name filter: matches the organization string anywhere in the certificate, which catches hosts behind a wildcard or SAN-only certificate that never names an individual domain.Infrastructure pivoting
Everything scanned in an organization's netblocksShodanorg:"<organization name>"Keyed on network registration data, which is the weak link: subsidiaries, acquisitions, and anything cloud-hosted usually sit under somebody else's org string entirely. Treat a clean result as "this is what is registered to you", not "this is all of you".Infrastructure pivoting
Everything scanned in a CIDR rangeShodannet:<CIDR>The precise version of the org: query above, and the one to use once you have the actual allocation out of RIR data rather than trusting a name match.Infrastructure pivoting
Hosts sharing a faviconShodanhttp.favicon.hash:<hash>Shodan stores a hash of each host's favicon. Two servers built from the same panel or kit serve the same icon, so the hash still matches after the domain, certificate, and IP have all been rotated: one of the few pivots that survives an operator rebuilding.Infrastructure pivoting
Hosts sharing a TLS stack fingerprintShodanssl.jarm:<jarm hash>JARM fingerprints how a server's TLS stack answers a fixed set of probes. Identical software and configuration produce identical fingerprints, which is what groups C2 servers of one family together across hosting providers. Expect false positives from anything running a stock stack.Infrastructure pivoting
Hosts resolving under a nameCensys Platformhost.dns.names:"<domain>"The second opinion on Shodan's view. The two scan on different schedules from different vantage points, and disagreement between them is itself information, usually about something that recently appeared or recently went away.Infrastructure pivoting
Certificates issued to an organizationCensys Platformcert.parsed.subject.organization:"<organization name>"Censys keeps certificates as their own dataset, so this searches issuance records directly rather than only the hosts currently presenting them. Good for finding the certificates behind assets that are already offline.Infrastructure pivoting
Certificates by subject common nameCensys Platformcert.parsed.subject.common_name:"<domain>"The certificate-dataset equivalent of Shodan's ssl.cert.subject.cn filter. Run both: the two services log from different CT log sets and different scan histories, and neither is complete on its own.Infrastructure pivoting
Hosts that redirect to a domainCensys Platformhost.services.endpoints.http.redirect_chain.hostname="<domain>"Hosts whose HTTP redirect chain lands on this domain. Catches parked lookalikes, forgotten vanity domains, and traffic-laundering hops that point at you but carry none of your names in their own certificates.Infrastructure pivoting
Hosts in an ASN running a given serviceCensys Platformhost.autonomous_system.asn=<number> and host.services.port=443CenQL's `and` combines conditions across the whole host record. Swap the port for whatever you are chasing; the ASN keeps it to one network rather than the whole internet.Infrastructure pivoting
A specific service on a specific portCensys Platformhost.services: (software.product="<product>" and endpoints.http.html_title="<title>")The nested form: parentheses after `host.services:` require both conditions to hold for the same service object, rather than for the host as a whole. Without the nesting you get hosts that run the product somewhere and show that title somewhere else.Infrastructure pivoting
Every certificate ever logged for a domain's subdomainscrt.sh%.<domain>The best free subdomain inventory there is, because certificate transparency records issuance rather than reachability. Hosts that were never linked, never crawled, and are long since decommissioned are all still in here.Infrastructure pivoting
Certificates naming an organizationcrt.sh<organization name>The identity search also matches organization names in certificate subjects, which surfaces assets under domains you did not know to ask about: the usual way a forgotten acquisition's infrastructure turns up.Infrastructure pivoting
Public scans of a domainurlscan.iopage.domain:"<domain>"Everything anyone has publicly scanned on this domain, with the full request chain and a screenshot per scan. Often the only surviving record of what a page looked like before it was pulled.Infrastructure pivoting
Where a domain appears as a subresourceurlscan.iodomain:"<domain>" AND NOT page.domain:"<domain>"Scans that contacted this domain without it being the page being scanned. Read one way it is your third-party dependency footprint on other people's sites; read the other way it is every scanned page that reached out to infrastructure you are investigating.Infrastructure pivoting
Recent scans of an IPurlscan.iopage.ip:"<ip>" AND date:>now-30dTies an address to the hostnames actually served from it recently. `date:>now-30d` is the documented relative-date form and keeps a long-lived shared-hosting IP from burying the thing you are after.Infrastructure pivoting
Every URL ever captured under a domainWayback Machine (CDX API)https://web.archive.org/cdx/search/cdx?url=<domain>/*&collapse=urlkey&fl=original&output=jsonReturns every distinct URL the Wayback Machine has captured under a domain, retired endpoints and removed pages included. `collapse=urlkey` dedupes to one row per unique path instead of one per snapshot, which is what makes this usable as a URL inventory rather than a crawl log.Infrastructure pivoting
Domain registration dataRDAPhttps://rdap.org/domain/<domain>rdap.org auto-redirects to the correct registry's RDAP server, so one URL works for any TLD without having to know which registry to ask. Returns registrar, registration/expiry dates, and nameservers as structured JSON.Infrastructure pivoting
IP allocation dataRDAPhttps://rdap.org/ip/<ip>The IP equivalent of the domain lookup above: which RIR allocation an address belongs to and who it is registered to, the structured-data version of what net:/org: on Shodan assumes you already know.Infrastructure pivoting
Is this IP internet background noiseGreyNoise Community<ip>Before spending time chasing a scanning source in your logs, check whether GreyNoise already classifies it as mass-scanning infrastructure everyone sees. A "benign"/"malicious" classification with no further digging is often the fastest triage decision available.Infrastructure pivoting
Lookalike domains carrying a brand stringcrt.sh%<brand>%Substring wildcards on both sides match the brand anywhere in a certificate identity, which is how you find brand-alike registrations the moment they request a certificate, typically hours to days before the page goes live and long before anyone reports it.Phishing & malicious domains
Certificates for a suspect domain's subdomainscrt.sh%.<suspect domain>Once you have one confirmed lookalike, this enumerates what else was issued under it. Phishing operators reuse a registration across several targets, so the sibling names usually name the rest of the campaign's victims.Phishing & malicious domains
Newly registered domains scanned recentlyurlscan.iopage.domain:"<domain>" AND page.domainAgeDays:<30Domain age is a strong phishing signal on its own. Legitimate business infrastructure is rarely three weeks old. Both halves are documented urlscan fields; combine with a brand term to scope it.Phishing & malicious domains
Community-tagged phishing scansurlscan.iotask.tags:"phishing" AND page.domain:"<domain>"Submitters tag scans, and phishing is the most consistently applied tag on the platform. Cheap corroboration when you are deciding whether something is worth escalating.Phishing & malicious domains
Pages by certificate issuer and titleurlscan.iopage.tlsIssuer:"<issuer>" AND page.title:"<brand>"Kits get deployed with the same free-CA certificate and the same copied page title over and over. Pairing the two is a surprisingly durable signature for one operator's output.Phishing & malicious domains
Hosts serving a string from a confirmed kitShodanhttp.html:"<distinctive string>"Matches a literal string in the page body across everything Shodan has crawled. Take a genuinely distinctive artifact out of a confirmed phishing page (a form action, a misspelled label, a hardcoded element id), and this finds every other host running the same kit.Phishing & malicious domains
Hosts titled after a brandShodanhttp.title:"<brand>"Impersonation pages copy the real page title verbatim, because that is what makes the browser tab look right. Noisy on its own; useful once you narrow it with a port, a country, or a certificate filter.Phishing & malicious domains
The kit's own sourceGitHub code search"<distinctive string>" NOT is:forkPhishing kits are shared as source far more often than people expect. Feeding a distinctive string from a page into code search sometimes lands you the whole kit, including the exfiltration endpoint the operator forgot to change.Phishing & malicious domains
URL already flagged as maliciousabuse.chhttps://urlhaus.abuse.ch/browse.php?search=<domain>URLhaus is a crowd-fed feed of URLs actively distributing malware. A hit means this exact URL or domain is already known and tagged, which is corroboration you did not have to produce yourself.Phishing & malicious domains
Samples sharing an import hashVirusTotal Intelligenceentity:file imphash:"<imphash>" p:5+The import hash summarizes a PE's import table, which is a property of how it was compiled rather than of its contents, so it survives repacking and recompilation that change the file hash completely. `p:5+` trims the noise floor to samples at least five engines flag.File, hash & sample hunting
Fuzzy-hash neighborsVirusTotal Intelligenceentity:file ssdeep:"<ssdeep hash>"ssdeep is a context-triggered piecewise hash: similar files produce similar hashes. Use it when you suspect you have one build out of a series and want the others.File, hash & sample hunting
Structural-similarity neighborsVirusTotal Intelligenceentity:file vhash:"<vhash>"VirusTotal's own structural clustering hash. It tends to group a family more tightly than ssdeep does, at the cost of missing more distant relatives. Run both and compare the two result sets.File, hash & sample hunting
A family within a time windowVirusTotal Intelligenceentity:file engines:"<family name>" fs:<YYYY-MM-DD>+Everything any engine labeled with this family name, first submitted after a date. `fs:` is first submission, so this reads as "when did this family start showing up", not "when was it last seen".File, hash & sample hunting
Samples signed by one identityVirusTotal Intelligenceentity:file sigcheck:"<signer name>" p:1+Once one abused or stolen code-signing certificate turns up in an investigation, this finds the rest of what was signed with it, usually the fastest route to the full scope of a supply-chain or signed-loader campaign.File, hash & sample hunting
A filename that keeps appearingVirusTotal Intelligenceentity:file name:"<filename>" p:3+Matches the names files were submitted under. Weak evidence on its own (names are trivially changed), but a distinctive dropped filename recurring across unrelated submitters is a real lead.File, hash & sample hunting
Where a file was served fromurlscan.iofiles.sha256:"<sha256>"urlscan records the files each scanned page served, so this turns a hash into the URLs and hosts that delivered it. Turns "I have the payload" into "I have the delivery chain".File, hash & sample hunting
A hash in public detection contentGitHub code search"<sha256>" NOT is:forkHashes land in Sigma rules, YARA files, and IOC lists well before they land in a blog post. `NOT is:fork` keeps the same rule repository from being returned three hundred times.File, hash & sample hunting
Hosts serving the same payload pathShodanhttp.html:"<staging filename>"Once you have a staging filename out of a sample, this finds the other open directories serving it, which is generally the rest of the campaign's infrastructure with no further work.File, hash & sample hunting
Samples by import hash (MalwareBazaar)abuse.chhttps://bazaar.abuse.ch/browse.php?search=imphash%3A<imphash>MalwareBazaar's free counterpart to the VirusTotal Intelligence imphash query above: no paid tier required, at the cost of a smaller corpus than VT's.File, hash & sample hunting
IOCs tagged to a family (ThreatFox)abuse.chhttps://threatfox.abuse.ch/browse/malware/<family>/ThreatFox indexes IOCs (IPs, domains, hashes, URLs) tagged to a specific malware family by the submitting analyst community, a fast way to pull a starter indicator set for a family you just got a name for.File, hash & sample hunting
Environment files in an organizationGitHub code searchorg:<github org> path:*.envA committed .env is the single highest-yield finding in this whole section. They exist to hold credentials, and they get committed constantly. `path:` takes globs, so this matches the file wherever it sits in the tree.Exposed & leaked assets
Private keys by file contentGitHub code searchorg:<github org> content:"BEGIN RSA PRIVATE KEY"`content:` restricts matching to file contents rather than paths or metadata: the thing no web search engine can do. The PEM header is a fixed string, so this has effectively no false-positive rate.Exposed & leaked assets
Terraform state filesGitHub code search"<domain>" path:*.tfstateTerraform state records the resources it manages along with, in older or carelessly configured setups, their secrets in plaintext. It also maps your cloud estate in one file, which is worth just as much to somebody scoping you.Exposed & leaked assets
Internal hostnames in live codeGitHub code search"<internal hostname>" NOT is:fork NOT is:archivedExcluding forks and archived repositories cuts most of the duplicate noise, leaving code somebody is plausibly still running. An internal hostname in a public repo maps your network for free.Exposed & leaked assets
CI workflow definitionsGitHub code search"<domain>" path:.github/workflowsWorkflow files describe the build pipeline, the registries it pushes to, and the names of every secret it consumes. Even with the values redacted, the shape of the pipeline is useful reconnaissance.Exposed & leaked assets
Private key files by extensionGitHub code searchorg:<github org> path:*.pemA .pem file in a public repo is a private key, a certificate, or both, and the extension alone is enough to find it without knowing what filename to look for. Combine with `content:"BEGIN"` to skip .pem files that only hold a public certificate.Exposed & leaked assets
Scoped to one languageGitHub code searchorg:<github org> "<internal hostname>" language:python`language:` narrows a content search to one language, which is the difference between a manageable result set and scrolling past every generated lockfile and vendored dependency that happens to mention the same string.Exposed & leaked assets
Screenshots of everything in your estateShodanorg:"<organization name>" has_screenshot:trueShodan screenshots services with a visual protocol: RDP, VNC, web panels. Scrolling the grid is genuinely the fastest way to see what your estate looks like from outside, and unattended desktops and default panels are impossible to miss.Exposed & leaked assets
Exposed remote desktopShodanorg:"<organization name>" port:3389RDP straight to the internet remains one of the most common initial-access paths in real incidents. This is the one query on this page worth running against your own org string before you finish reading it.Exposed & leaked assets
A specific product inside a rangeShodannet:<CIDR> product:"<product>"Scoped to a range you actually own, this answers "how many of these are exposed" without the ambiguity of matching on a registration name. Use it when a CVE lands and you need a number by the end of the meeting.Exposed & leaked assets
Vulnerable services in a networkCensys Platformhost.autonomous_system.asn=<number> and host.services.vulns.id:"<CVE id>"Censys attaches vulnerability identifiers to service records, so this scopes a specific CVE to a specific ASN. The equivalent Shodan filter (vuln:) is restricted to paid memberships, which makes this the free route to the same answer.Exposed & leaked assets
Shadow IT and forgotten subdomainscrt.sh%.<domain>Same query as the infrastructure track, different question. Read the result as an asset inventory: anything in here you cannot account for is either shadow IT, an abandoned project, or somebody else's certificate for your name, and all three are worth an answer.Exposed & leaked assets
Detection rules naming an actorGitHub code search"<actor name>" (path:*.yml OR path:*.yar) NOT is:forkSigma rules are YAML and YARA rules are .yar, so the two path globs cover most published detection content. The rule that catches a campaign is usually public before the report describing it is.Actor & campaign research
Proof-of-concept and exploitation toolingGitHub code search"<CVE id>" path:*.py NOT is:forkTells you how weaponised a vulnerability actually is, which is a different and more useful question than its CVSS score when you are deciding what to patch tonight.Actor & campaign research
Scans tagged for a campaignurlscan.iotask.tags:"<campaign tag>"Researchers tag their submissions, so a campaign tag often gathers scans from several unrelated people. Good corroboration, and occasionally a set of hosts nobody has published yet.Actor & campaign research
Panels by issuer and title togetherShodanssl.cert.issuer.cn:"<CA name>" http.title:"<panel title>"C2 panels are consistent about both the free CA they get certificates from and what their login page calls itself. Either filter alone is far too noisy; together they narrow to something reviewable by hand.Actor & campaign research
Where a family is being submitted fromVirusTotal Intelligenceentity:file engines:"<family name>" submitter:<ISO country code>`submitter:` takes an ISO 3166-1 alpha-2 country code. Submission geography is a rough proxy for targeting, and a family suddenly appearing from a new region is worth noticing.Actor & campaign research
A software and port combinationCensys Platformhost.services: (port=<port> and software.product="<product>")Once a report names the software and port a campaign's servers run, the nested form finds the rest of them. Requires both conditions on the same service object, which is what stops it matching unrelated hosts that happen to satisfy each separately.Actor & campaign research

Related

All references