Understand the report inside the attachment
Open DMARC XML, GZIP and ZIP Reports Locally
A DMARC attachment can be readable XML or a compressed container around XML. The extension describes packaging, not whether the report is authentic, complete, or useful for a particular investigation. Preserve the original privately, choose a supported input, and let the analyzer check the bounded structure before interpreting counts. A rejected file needs an explanation, not a series of undocumented edits until a green result appears. An accepted file still contains supplied declarations rather than independently verified mail activity.
The DMARC Report Analyzer offers Files and Paste XML modes. Select attachments through Report files, or enter XML text under XML report. Load example fills input only; choose Analyze reports to begin. Clear ends the working inspection. There is no mailbox connection, remote-file URL input, API submission, or report upload to an analysis service.
Preserve the original and identify the actual input
Obtain the report attachment through your authorized report-handling process and keep an unchanged original under your organization’s rules. Do not paste the surrounding email, its MIME wrapper, a screenshot, or a summary table into XML report. The analyzer needs the actual aggregate-report XML, either directly or inside a supported compressed file. Renaming a spreadsheet to .xml does not create XML, and renaming an archive does not change its compression format. Record collection context outside the tool if your investigation needs it. A filename and an apparent sender address are useful leads, but neither proves the reporting organization named inside the document actually generated it.
The accepted file extensions are .xml, .gz, and .zip. Files mode handles their packaging locally; Paste XML is for already available XML text, not compressed bytes or encoded attachment text. If you use another approved application to obtain the XML, keep that step separate from the analyzer and preserve the original attachment. Do not assume that a browser preview preserved every character when copying. Truncated text, missing closing elements, or material copied from a formatted email can produce a different document. A successful analysis of an edited derivative does not retroactively establish that it matches the original attachment byte for byte.
For collection details, use the relevant Google reports guide, Microsoft reports guide, or Yahoo reports guide. Keep provider-specific collection separate from file interpretation: this panel does not sign into those services, configure reporting, fetch missing attachments, or establish a report’s chain of custody. Only inspect reports you are authorized to handle. Aggregate reports can still disclose domains, source addresses, contact information, and operational patterns even though they are not full message bodies. Local processing does not make that material automatically suitable for a public ticket or screenshot.
Understand compressed packaging and strict ZIP acceptance
A GZIP file is decompressed locally before its XML is read, and both source and expanded limits apply. A small attachment can expand into a much larger document, so its download size alone cannot establish acceptance. GZIP integrity errors are rejected rather than ignored to recover a convenient fragment. The genuine analyzer uses native browser decompression for GZIP; it does not send the attachment to a remote conversion service. If processing fails, preserve the original and check whether your authorized collection process produced a complete attachment. Repeatedly renaming a damaged file or pasting a partially recovered payload cannot supply the missing integrity evidence.
A ZIP must contain exactly one XML file, with optional directory entries and nothing else. There may be no more than 64 entries in total, counting directories as well as the XML file. Two XML files are ambiguous and rejected, as is one XML plus a readme, image, unrelated document, or extra hidden file. An XML inside an ordinary directory path can satisfy the rule; a nested archive cannot. Encrypted ZIPs, additional files, nested archives, corrupt data, CRC failures, and inconsistent local and central metadata are rejected. The analyzer does not choose the first plausible XML and ignore the rest. These restrictions define this tool’s accepted report container, not a claim that every other ZIP layout is universally invalid or malicious.
ZIP reading is bounded in memory and does not extract archive entries to disk. A directory entry is not another report, but an additional regular file is not excused merely because its name looks administrative. If your legitimate bundle contains several reports, obtain separate authorized report files outside the analyzer while preserving the original bundle and noting that preparation step. Do not strip unexplained entries and call the resulting archive untouched evidence. There is no password prompt, nested-archive workflow, arbitrary archive browser, or remote codec download. A rejection is a reason to review the input and acceptance rules, not a verdict about the person or organization that sent the attachment.
Check UTF-8 and the namespace without rewriting meaning
XML input must be UTF-8; an optional UTF-8 byte order mark is supported. Another encoding is not made acceptable by changing only its declaration to say UTF-8. If an authorized workflow needs conversion, perform that separately, retain the original, and identify the converted file as a derivative. The analyzer is not a general encoding repair utility. Paste XML likewise needs complete report text rather than a visual rendering with omitted markup. Do not delete accented contact text, comments, or unfamiliar elements simply to make an error disappear. An undocumented edit can change meaningful report content and later affect comparison even if the remaining text becomes parseable.
The reader supports legacy feedback reports in an empty namespace, named legacy reports using http://dmarc.org/dmarc-xml/0.1, and RFC 9990 reports using urn:ietf:params:xml:ns:dmarc-2.0. Core children in a named legacy report may be qualified or unqualified. Correctly bound prefixes are supported as well as a default namespace; a prefix is not required to have a particular spelling. The namespace URI and binding determine which elements are recognized, not just a familiar local name. An unknown root namespace or namespace spoofing is rejected. Do not remove prefixes or replace a namespace URI to make an unrelated document resemble a report. That changes the document’s meaning instead of demonstrating that the original belonged to a supported format.
Foreign extension namespaces are ignored with an explicit warning rather than interpreted as core fields merely because their local names look familiar. Read that warning as a limit on what the summary includes. Current aggregate reporting is specified by RFC 9990 alongside the DMARC protocol in RFC 9989; RFC 7489 provides legacy context rather than the sole current specification. This analyzer remains a syntactic, bounded structural reader, not a full XSD-conformance certification service for either generation. A successful result means supported content passed this reader’s checks. It does not certify every optional construct, ignored extension, protocol requirement, or operational assertion inside the supplied report.
Start with a supported original
Open report files locally
Choose XML, GZIP or a supported single-report ZIP, then explicitly start analysis inside the panel.
Apply every size, structure and execution limit together
Select no more than 10 files. Each source file is limited to 4 MiB, and the batch source total is limited to 16 MiB. Each expanded XML is also limited to 4 MiB, with no more than 16 MiB expanded XML across the batch. MiB refers to binary units, so 4 MiB is 4,194,304 bytes and 16 MiB is 16,777,216 bytes. Compression does not bypass the expanded ceiling, and choosing fewer files does not raise the per-file ceiling. Pasted XML must also fit the XML boundary. Check both the size of the selected attachment and the size of its actual report content when diagnosing an oversized-input rejection.
Structural ceilings apply in addition to byte limits: at most 5,000 rows per report, 10,000 rows per batch, XML depth 24, 150,000 nodes, and 4,096 UTF-16 units in any leaf text value, comment, attribute value or element local name. Elements, attributes, text events and comments contribute to the shared parser-node ceiling. A compact document can exceed a row or node limit before reaching its byte ceiling. A huge single text value can fail even when the report contains few rows. All limits apply together; passing one check is not an exemption from the others. Do not delete rows or truncate text to manufacture a complete-looking analysis. If authorized preparation creates a limited derivative, label it honestly and never compare its reduced counts as though they represented the untouched original report.
DTD and entity declarations and external processing instructions are rejected; there are no external entity calls or remote schema fetches. Malformed XML is rejected rather than repaired into a guessed report. A local outer worker enforces a 12-second deadline across the analysis, including decompression and parsing. The batch succeeds atomically: corrupt input, a structural violation, a limit failure, or a report conflict prevents partial success from being published. Do not interpret a failed batch as showing zero instances or assume that early files were accepted for comparison. The time limit bounds this working operation; it is not a promise that every otherwise supported report will finish on every device.
Read the accepted report without inventing missing facts
A usable report needs its declared organization, contact, report ID, begin and end times, and published policy domain and policy value. Timestamps must be safe integer epoch values within the supported UTC range, with beginning no later than ending. The analyzer does not infer a missing window from a filename or the date you opened an email. Rows need a valid four-part decimal IPv4 or IPv6 source address, a nonnegative safe integer count, and an author domain in header_from. These checks concern usable structure. They do not establish that the IP operator, named organization, or domain supplied accurate evidence or authorized any underlying message.
After analysis, read the statistics in their report context, use the report filter, inspect source-IP groups, and open row details. The plain raw XML viewer shows the selected report as text, not rendered active markup. Raw identifiers, authentication results, override reasons, and policy options remain supplied declarations. Preserve distinctions between aligned SPF and DKIM in policy_evaluated and raw checks under auth_results. The derived outcome passes if either aligned method passes, fails only when both fail, and is otherwise unknown. See the alignment guide when a raw pass and a reported alignment failure appear together; they are not necessarily contradictory observations.
Counts weight rows as reported message instances, not unique emails, sent mail, or proof of delivery. Disposition remains separate from alignment. The comparison guide explains same-key meaningful duplicates, explicit skip counts, conflicting content that aborts the batch, and overlap warnings that do not deduplicate messages. The failure investigation guide explains matching sources to an approved sender inventory without declaring unfamiliar IPs hostile. For a particular delivered message, the Email Header Analyzer is a separate single-message reading tool, not an automatic join between headers and aggregate counts or a replacement for original receiver evidence.
Handle the working session and external originals separately
The isolated analyzer processes files and pasted text locally, without API requests, mailbox access, or report uploads. It performs no domain DNS lookup, SPF evaluation, DKIM cryptographic verification, or fresh DMARC validation. It also supplies no ASN, PTR, reputation, or geolocation enrichment for source addresses. The standalone panel uses no-store behavior and blocked outbound connections; ordinary delivery of the surrounding website is a separate activity. These constraints describe how the genuine panel handles a bounded reading task, not a guarantee that a supplied report is authentic or that your entire browser and device are trustworthy. Keep operational conclusions attached to the evidence actually available.
Clear stops the worker and removes the file input selection, pasted XML, statistics, report filter, IP and row details, plain raw XML viewer, and every derived result. A hidden page, pagehide, pageshow, or freeze also clears that working state, so returning to the page does not restore a previous report. The original attachment remains wherever you kept it outside the panel. Cleanup is not secure memory erasure and cannot promise protection against browser extensions, a compromised device, screenshots, or a hostile parent replacing the interface. Avoid claiming that Clear deletes the downloaded file or controls unrelated applications and their retained copies.
There is no Copy button, export, PDF, project, history, persistent storage, or automated redaction. Any external working notes, manually prepared derivative, or disclosure require your own approved handling process. Before sharing a screenshot or excerpt, review contact fields, report IDs, domains, addresses, and comments for unnecessary exposure. Keep originals private when a synthetic example would answer the question. Resolve unsupported or damaged inputs with the authorized source rather than silently repairing them into supposed evidence. Return to the DMARC Report Analyzer overview for another local reading task, or browse Tools without carrying report data into navigation links.
Questions about DMARC XML, GZIP and ZIP Files
Can one ZIP contain several XML reports or an extra readme?
No. A ZIP must have exactly one XML file plus optional directories, with at most 64 entries total including those directories. Additional files, nested archives, encryption, corruption, CRC failures and inconsistent archive metadata are rejected. Keep a legitimate multi-report bundle privately and prepare separate authorized report files outside this tool if needed.
Are prefixed RFC 9990 reports supported?
Yes, when the prefixes are correctly bound to urn:ietf:params:xml:ns:dmarc-2.0. Legacy empty-namespace reports and named legacy http://dmarc.org/dmarc-xml/0.1 reports are also supported, including qualified or unqualified legacy children. Unknown root namespaces and spoofing are rejected; foreign extensions are ignored with a warning. Parsing is not full XSD certification, and removing prefixes is not an appropriate repair.
Does a small compressed file avoid the XML limits?
No. Source size, expanded size, batch totals, row counts, depth, nodes, leaf text and the 12-second worker deadline all apply together. A rejected file makes the batch fail without partial successful results. Compression changes packaging, not the permitted size or complexity of the XML report.
Continue with another DMARC-report guide
Read an aggregate report in the DMARC Report Analyzer & XML Viewer or choose a related task below. Explore more practical utilities in all free tools.
Read the document within its limits
Inspect a supported DMARC report
Review supplied report structure and declarations locally, then clear the working input and results.