Keep a verifiable diagnosis without sharing the file
When you run diagnosis, you can download a local JSON report containing the detected format, the structural findings, the original file size and its SHA-256 digest. The report does not embed the original document. If a trusted SHA-256 was supplied independently, a mismatch blocks processing before conversion or repair. You can paste the digest or import a single-record .sha256 file whose filename matches the original. After a successful transformation, the output has its own downloadable SHA-256 record, checked again against the bytes before it is offered. Matching hashes establish byte identity with the chosen reference, not malware safety or publisher authenticity.
Correct an extension without changing document bytes
Use Diagnose only even when a conversion request is already typed. If a supported file has the wrong extension, File Doctor can offer a copy under the detected format’s extension (for example, a valid DOCX saved as .doc, or a PNG saved as .jpg). It verifies the original SHA-256, runs the normal output-format checks under the corrected name, and checks the SHA-256 again before download. You can also export a .sha256 record for the corrected filename. No bytes are converted, reconstructed or silently discarded. Damaged, unsupported or unverified containers are not approved as successfully renamed files; PDF page validation still requires the full local PDF parser.
What this diagnostic checks now
MakeFileWork identifies JPG, PNG, WebP, GIF, PDF, ZIP and Office Open XML containers such as DOCX, XLSX and PPTX. It flags filename-extension mismatches, validates image dimensions and common end markers, checks supported ZIP/Office containers against their central directory, local headers and CRC32; unsupported or oversized archives are not marked verified, can conservatively rebuild a missing ZIP directory from intact local records, and performs conservative PDF envelope repair when possible.
How Office files are distinguished from ordinary ZIP archives
DOCX, XLSX and PPTX are ZIP containers, but a ZIP is not an Office document merely because its data or comment mentions word/document.xml. Detection now examines actual ZIP directory entries and their local headers for [Content_Types].xml, _rels/.rels and the correct Word, Excel or PowerPoint root part. A damaged directory can sometimes be recognized from intact consecutive local records for recovery; the later Office integrity inspection, not filename detection, determines whether the document is usable.
Why apparently harmless archive filenames may be unsafe
ZIP and TAR entries can contain names that look ordinary but change on another operating system. For example, ＣＯＮ.txt becomes the reserved Windows name CON.txt after Unicode normalization; full-width punctuation may become a colon, and percent-encoded sequences can be decoded by downstream tools. Archive diagnosis distinguishes a valid checksum from safe extraction: a checksum alone cannot approve an ambiguous filename. If such an entry is flagged, inspect the archive and request a version with safe names rather than forcing extraction.
When ZIP and Office signatures disagree
A normal ZIP may contain Word and Excel entry names at the same time, but that does not make it a DOCX or XLSX. MakeFileWork checks the complete directory before choosing an Office workflow; it also compares each required entry with its local header, including compression method, flags, checksum and compressed size where recorded. Contradictions stay on the ZIP inspection route rather than being presented as a verified Office file. Inspect the archive to review its structure before deciding whether to attempt Office repair.
If the application already gives you a specific error, go straight to the matching recovery workflow: Word found unreadable content, Excel cannot open the file, PowerPoint found a problem with content, or PDF failed to load or is damaged. Those pages use the same universal assistant but start with the right repair intent and explain the evidence MakeFileWork can verify.
Why a ZIP end marker alone is not enough
A ZIP may contain the four-byte end-of-directory signature even when its central directory points outside the file, a local entry disagrees with its directory record, or its contents fail CRC32. For supported archives up to 40 MB, diagnosis checks those structures and each entry’s checksum. A matching filename extension or a visible ZIP signature is never sufficient evidence of successful verification.
Warnings, encrypted archives and safe batch output
Encrypted entries, unsupported ZIP features, suspicious paths and files beyond the deep-verification limit are not certified as intact. A warning means the tool could not prove integrity, not necessarily that the file is corrupt. For a downloadable mixed-file batch, an unverified diagnostic result is rejected rather than silently labeled verified. Use archive inspection for more detail, and verify the contents of Office files in their intended application.
Text encoding checks are bounded
The diagnostic examines up to 1 MiB of text. It handles a UTF-8 character split at the sample boundary without declaring the file malformed, but reports a warning when later bytes remain uninspected. That is not a full-file encoding certificate; choose a dedicated text repair workflow if the rest of the file may be damaged.
What it does not claim
A file can still contain damaged application data even when its outer structure looks valid. MakeFileWork reports only the checks it actually performed. It does not invent missing bytes, bypass unknown passwords or claim a repair succeeded when recovery cannot be verified.
Why a correct file can still fail to open
Common causes include a wrong extension, an interrupted download, a truncated ending, a damaged ZIP central directory, an incomplete image marker or corruption inside an Office package. The diagnostic is designed to separate those cases before you choose a more specific repair.
PDF, TAR and GZIP: what can actually be verified?
A PDF header or %%EOF marker does not prove that every page can be opened. Diagnosis checks the PDF envelope and cross-reference pointer, then returns a warning until a full PDF parser checks pages and embedded resources. TAR diagnosis uses the existing archive reader to check header checksums, entry boundaries, padding, safe paths and the end blocks. GZIP diagnosis actually decompresses the stream within a 100 MB expansion limit and verifies its checksum; a .tar.gz also needs valid TAR entries. A corrupt checksum, missing PDF cross-reference or truncated archive is an error, not a successful inspection. Use TAR/GZIP extraction only after the relevant checks.
Why a GIF end marker is not enough
A final GIF trailer byte alone cannot prove that frames, palettes or compressed image data are intact. The diagnostic reports an unverified warning until a complete decoder checks the frames; a missing trailer is an error. Do not treat the signature as proof that the image is viewable.
JPEG, WebP and GIF: a matching file ending does not mean the image is decoded
A JPEG can contain a start marker, width, height and final marker without any actual image scan. The diagnostic now checks the marker chain and requires an SOS scan header; a missing scan is a structural error. A real scan is still only a warning because the compressed pixels are not decoded by this check. If a JPEG needs a different format, use JPG to PNG only after confirming it opens correctly.
For WebP, the RIFF size must match the file, every chunk must fit its declared boundary and at least one image frame must be present. An empty VP8X canvas is not a valid picture. For GIF, the parser walks color tables, image descriptors, extension sub-blocks and the trailer; a trailer after arbitrary bytes is rejected. These checks have a 40 MB limit and do not decode VP8 or LZW image pixels. If the image opens successfully, WebP to PNG or GIF to PNG may help with compatibility; conversion cannot reconstruct missing image data.
Why a valid ZIP checksum is not enough for Office files
A DOCX, XLSX or PPTX is a ZIP package of connected parts. Its ZIP directory and CRC32 checks can pass even when an embedded image, worksheet or slide referenced by the document is missing. File Doctor now checks required Office package parts and internal references as well as ZIP integrity. In independent examples, a Word document with a missing picture, an Excel workbook with a missing sheet and a PowerPoint with a missing slide were all correctly marked as unverified even though their ZIP CRC32 checks passed. These are local structural checks, not proof that Word, Excel or PowerPoint can open the file. Try Office repair for a cautious recovery; missing original content cannot be invented.
Privacy
These diagnostic checks run in your browser. Your file is not uploaded by this workflow.