Paste a header, read the story
Every email carries the record of where it came from and what each server concluded. Paste it on the left, press V to paste from the clipboard, or drop an .eml or .msg file here.
header-analyzer [overview|authentication|path|findings|filters|headers] headerEvery email carries the record of where it came from and what each server concluded. Paste it on the left, press V to paste from the clipboard, or drop an .eml or .msg file here.
Everything is worked out in this tab. Only your display settings are saved in the browser, never the header.
Open the message in your mail client and copy its header: Show original in Gmail, View message details in Outlook on the web, File → Properties in Outlook for Windows. Or save the message as an .eml or .msg file.
Paste it into the box, press V anywhere on the page, or drop the file on the window. The analysis starts at once and nothing leaves the tab.
The ribbon shows a score out of 100, the SPF, DKIM and DMARC results and whether a passing check aligns with the From address. Overview explains it in plain English.
Open Path to see every hop, oldest first, with the time spent at each, the TLS version, the sending IP and its reverse DNS. Click a server on the route to jump to its Received line.
Switch on Hide personal details, then copy a summary for a ticket, save a report or JSON, print to PDF, or copy the header with addresses and the subject masked.
From?.msgA message header analyser reads the raw technical lines at the top of an email and turns them into something you can act on. Every server that touches a message stamps a Received line onto it, and the receiving system records what SPF, DKIM and DMARC concluded. Reading that by hand is slow and easy to get wrong. This tool rebuilds the delivery path, works out whether the authentication really proves the sender is who they claim to be, and points at the patterns that mean spoofing or phishing. It answers to both spellings, analyser and analyzer.
It runs entirely in your browser. A header carries internal hostnames, private IP addresses, recipients, subject lines and ticket references, and many online analysers send all of that to their own servers. Here there is no server to send it to, which matters when the message is a customer's phishing report or part of an internal incident.
SPF, DKIM and DMARC with real alignment, relaxed and strict, the signing domain of every DKIM signature, an estimate of the key size and the ARC chain.
Every hop oldest first, the time spent at each server, TLS version and cipher, the first public IP and its reverse DNS, drawn as a route you can click through.
More than fifty checks: display-name tricks, lookalike and punycode domains, duplicate From headers, Reply-To and Return-Path mismatches, odd dates and broken chains.
Microsoft 365 X-Forefront-Antispam-Report decoded into SCL, BCL, PCL, SFV and category, and SpamAssassin scores with the meaning of each rule.
In Gmail, open the message, choose the three-dot menu and Show original, then copy to clipboard; the summary box above the header is ignored. In Outlook on the web use the three-dot menu, View, View message details. In Outlook for Windows open the message in its own window and choose File, Properties, then copy the Internet headers box, or save the message and drop the .msg file on this page. Apple Mail has View, Message, All Headers, and Thunderbird shows the source with Ctrl+U. Copy from the top down to the first blank line: that line is where the header ends and the body begins, and anything after it is ignored.
Headers have three kinds of content. What the sender claimed is From, To, Subject, Date and Reply-To, all supplied by whoever wrote the message and therefore all forgeable. What the infrastructure recorded is the Received chain, Authentication-Results and Received-SPF. What the filters concluded is vendor headers such as X-Forefront-Antispam-Report or X-Spam-Status. The value of a header is almost all in the second and third groups.
Each server adds its Received line to the top of the message, so the stack is in reverse order: the bottom line is the first hop, closest to the sender, and the top line is the last, closest to you. This analyser reverses them and numbers them in the order the message travelled, so hop 1 is the start. It shows the delay between hops, which separates a slow relay from greylisting: a single large delay at one hop with normal timings either side is the classic shape of a first message from an unknown sender, retried minutes later. A hop over a minute usually means queuing; over an hour usually means the receiving server deferred the message and tried again.
You can trust the top of the chain, which your own infrastructure wrote, and should treat everything below the first server you control as a claim. An attacker cannot forge the lines added after the message left their hands, nor produce a DKIM signature that validates for a domain whose private key they do not hold. A fabricated hop shows as a break in the chain, which is flagged.
SPF publishes in DNS which servers may send mail for a domain, and the receiver checks the connecting IP against that list. DKIM attaches a cryptographic signature so the receiver can verify the message was not altered and came from a domain holding the matching key. DMARC ties both to the visible From address and tells receivers what to do when neither lines up. SPF proves a server was allowed to send, DKIM proves the message is intact and signed, and only DMARC connects either fact to the name the recipient actually sees.
The most misunderstood part is alignment. DMARC does not pass because SPF or DKIM passed. It passes only if a passing check is also aligned with the domain in the From header. SPF checks the envelope sender in Return-Path, which is often a bulk platform's domain, and DKIM checks the d= domain of the signature. If neither matches the From domain, DMARC fails even though both report pass. Relaxed alignment accepts a matching organisational domain, so mail.example.com matches example.com; strict alignment needs an exact match. The Authentication tab lays this out as a ladder and names the mechanism that carried the pass.
That is also why SPF can pass on a fake email. SPF validates the Return-Path, not the From address your client shows. An attacker registers their own domain, publishes a valid SPF record, sends from an authorised server and puts anything they like in From. SPF passes for their domain and the recipient sees a spoofed name. A bare SPF pass proves very little on its own.
A few smaller points. dkim=none means no signature at all, which is different from dkim=fail, where someone signed and the signature did not validate. A message can carry several DKIM signatures from the platform, the customer and a mailing list; what matters is whether any valid one is aligned. The DMARC policy p=none only monitors, p=quarantine asks for junk-folder treatment and p=reject asks for refusal, and a domain that stays on p=none gets reports but no protection.
Forwarding breaks authentication by design. The forwarding server sends from its own IP, so SPF for the original domain fails, and any change to the message, such as a footer or a subject tag, breaks the DKIM signature. With both broken, DMARC fails on a perfectly genuine message. ARC, the Authenticated Received Chain, lets each relay record, with its own seal, the results it saw before it made changes. If a message failed DMARC but carries an intact ARC chain from a service you trust, that is usually forwarding rather than an attack. The tool looks for ARC and for forwarding patterns and says so, instead of reporting a bare failure.
Display-name spoofing puts something trustworthy in the part of From your client shows in large letters, such as a brand name, while the real address is controlled by the attacker and often hidden on a phone. A variant puts a whole fake address in the display name. The analyser checks display names for embedded addresses and for brands that do not match the sending domain.
A lookalike domain is registered to be confusable with a real one: rn for m, 1 for l, a zero for the letter O, or a Cyrillic character that looks identical to a Latin one, which appears in headers as punycode starting with xn--. These are flagged, as are mixed scripts.
Almost nothing legitimate sends two From headers. Gateways and mail clients disagree about which to show, and attackers use that: one is authenticated and the other is displayed. A duplicate of a header that should be unique is treated as a high-severity finding. A Reply-To on another domain sends your answer somewhere other than the apparent sender; there are legitimate uses, but it is also the standard mechanism of business email compromise, where a supplier's address is spoofed and replies are routed to the attacker. A Return-Path that differs from From is normal for newsletters and only meaningful when nothing else ties the message to the From domain.
In Microsoft 365, the X-Forefront-Antispam-Report header holds the verdict. SCL is the spam confidence level: -1 means allow-listed, 0 and 1 are not spam, 5 and 6 are spam and 9 is high-confidence spam. BCL is the bulk complaint level from 0 to 9, and a high BCL with a low SCL is typically legitimate but unwanted marketing. PCL is the phishing confidence level, and SFV records why the verdict was reached, for example SPM for spam filtering or SKA for allow-listed. These are decoded on the Filters tab.
SpamAssassin records a numeric total and the rules that fired in X-Spam-Status. The default threshold is 5.0, but the useful part is the rule names: SPF_FAIL is an authentication problem, BAYES_99 means the content resembles known spam, and URIBL_BLOCKED means a link was on a blocklist. Filter headers are added by the system that receives a message, so a header copied from a sent item will not have them.
The score starts at 100 and loses points for every finding, with authentication and impersonation weighing the most. It is also capped by the worst finding: one critical finding caps it at 40, a high one at 65 and a medium one at 85, so a single serious problem cannot hide among many clean results. The Findings tab shows each penalty so you can see how the number was reached and disagree with it. Treat it as triage, not a ruling. A forwarded message can score badly for innocent reasons, and a targeted attack from a correctly configured, compromised mailbox passes every check here. Headers describe transport and authentication; they say nothing about whether the content is a scam.
Drop a .eml file on the page and the header is read straight out of it; the body is left out. Outlook .msg files are read too: the tool opens the compound file in your browser and extracts the stored transport headers. Outlook keeps those only for mail that arrived over the internet, so a sent item or a draft has none and the tool says so. Plain text, .mbox and a pasted Gmail Show original page work as well. Nothing you drop is uploaded: the file is read locally by the browser.
The Share tab copies a plain-text summary, Markdown for a ticket, a text report or JSON, or prints a report to PDF. Turn on Hide personal details and the part of every address before the @, private IP addresses, internal hostnames and the subject are masked on screen and in everything you share. Header, masked copies the original header with those parts hidden, so you can ask a vendor or a forum for help without handing over what is inside.
A browser cannot query DNS or blocklists, and the only workaround is a server in the middle, which would mean uploading your header. So this tool does not look up SPF, DKIM or DMARC records itself and reports what the receiving server recorded. The Path tab gives one-click links with the sending IP and domain filled in for Spamhaus, MXToolbox, AbuseIPDB, Cisco Talos, a DNS lookup and WHOIS; they open another site, and only when you click. The alignment check uses an approximate public-suffix list, and the server's own DMARC result is the authority. Once the page has loaded the analyser needs no network at all, so it works on an isolated machine.
The header you paste or open is processed in this page and is not stored: it is gone when you close or reload the tab. The page keeps only your display settings in this browser's local storage, under the key jb-header-ui (time zone, the privacy switch and a few toggles) and jb-theme (light or dark). It makes no fetch or upload calls. The only outside requests are the JetBrains Mono font from Google Fonts and an anonymous, cookie-free page-view counter from SimpleAnalytics, and the lookup links on the Path tab, which open only when you click them.
To fix what the header shows, build the records with the SPF, DKIM and DMARC checker and record builder. For related work try the log explorer, the Conditional Access simulator, the NIS2 checker or the CVE explorer.