Local file safety check
Scan files before you open them or feed them to AI.
ScanBeforeOpen analyzes files locally in your browser for known risk patterns. No uploads, no accounts, no trackers.
Checks known risk patterns. Not a replacement for antivirus.
Ready
Drop file here
Save it. Do not open it. Scan first.
Maximum 50 MiB. Large files can take longer.
PDF, images, CSV, text, ZIP and Office, each with limited inspection scope. One file per scan. No upload.
Preview first, decide calmly
A file does not need to be opened immediately.
ScanBeforeOpen turns the quick double-click into a calm checkpoint: scan, understand, preview in a controlled way, and only open normally when source and content make sense.
What you get immediately
- A clear recommendation without panic language
- An isolated preview for suitable file types
- Clean exports where they are technically sensible
- Transparent limits instead of false safety promises
How it works
1
Select a file
Save the file first instead of opening it directly.
2
Check it locally
The check runs in your browser. The file is not uploaded.
3
Understand risks
You get a plain-language recommendation first, with details available below.
4
Use safe preview or a cleaned copy
Use a safe preview or, when available, a cleaned copy with reduced risk.
File types with limited inspection
Inspect locally now
Free tools without uploading. All use the same bounded scanner engine.
PDF risk checker: inspect a PDF locally
Choose a PDF to look for known JavaScript, action and attachment patterns in its raw data. Inspection stays in your browser. It is not a complete PDF or malware analysis.
ZIP risk checker: inspect before extracting
Inspect paths, extensions and size ratios in a ZIP before extracting it. No contained program is launched. The report does not establish the safety of its contents.
CSV injection checker: inspect formula prefixes
Check a UTF-8 CSV for cell prefixes that spreadsheet applications may interpret as formulas. The scanner identifies suspicious cells and offers a complete marked CSV within its budgets.
Privacy rebuild
Hard-to-detect fingerprints? Rebuild instead of guessing.
ScanBeforeOpen flags selected text, PDF and container signals. A rebuild creates a new bounded representation instead of repairing the original. Text excerpts, page images and single-frame PNGs have separate limits and can change content or functionality.
Available rebuild modes and limits
- Text rebuild: normalizes Unicode and removes invisible markers
- PDF visual flatten: renders up to twelve pages as images in a ZIP; larger documents are blocked
- Office data export: blocked pending reliable integrity validation
- Image canvas rebuild: creates a new PNG without typical container metadata
Typical original container and object traces can be omitted. Visible private details and watermarks can remain. No export guarantees anonymity or replaces legal or organizational approval.
Checks by file type
How ScanBeforeOpen examines files
Each file type has different risks. ScanBeforeOpen therefore uses matching local checks, previews, and clean-export options instead of treating every file the same.
Looks for JavaScript, OpenAction, Launch, attachments, forms, links, old PDF objects, document IDs, hidden layers, and prompt risks. Raw patterns only, not full inspection of compressed or encrypted objects. Canvas export up to twelve pages.
Office
Checks macro containers, external relationships, data connections, OLE/ActiveX, comments, metadata, customXml, docProps, and hidden sheets. Preview shows limited text excerpts. Office exports are blocked.
ZIP
Checks path traversal, risky extensions, nested archives, encryption, entry counts, and compression ratios.
CSV
Detects formula-injection patterns in cells and can export risky formulas as text.
Images
Checks image type, metadata, and container hints. Canvas rebuild can reduce typical metadata and many container traces, with possible format or quality changes.
SVG/HTML/Text
Looks for active content, external references, invisible characters, and prompt instructions. Preview stays as source or sanitized text.
What ScanBeforeOpen checks
Human vs. Machine View
Shows whether filenames, extensions, or invisible characters look different to people than to software.
Safe Preview
Displays content in a controlled way without running links, macros, or active content from the original file.
Clean Export
Creates local cleaned copies only where a matching export is available. This reduces risk, but it is not a guarantee.
Review text for AI
Surfaces known prompt patterns and removes invisible characters from text excerpts. Does not reliably prevent prompt injection.
Email and messenger help
Helps with attachments from email, chat, and downloads: save first, do not open, then scan.
Privacy by design
No file uploads, no analytics, no accounts, and no active advertising scripts.
Guides for safer files
Practical defensive guidance for attachments, downloads, Office files, archives, images, and AI workflows.
Check a PDF before opening: JavaScript, actions and limits
Inspect a suspicious PDF locally: understand JavaScript patterns, attachments and metadata signals, with a real test example and explicit raw-inspection limits.
Check a ZIP before extracting: paths, extensions and ZIP bombs
Understand ZIP contents before extraction: risky names, path traversal and compression limits, with local inspection and selective export explained.
Detect file type spoofing: extension, MIME type and magic bytes
Spot double extensions and wrong file types: what extension, MIME type, magic bytes and structure say, and why a recognized type is still no safety clearance.
Check files before sharing: metadata, hidden content, checklist
What a PDF, Office file or photo gives away when passed on: metadata, comments, hidden content. A practical checklist with the limits of ScanBeforeOpen.
Check an SVG file: scripts, external resources and a safe view
An unknown SVG can carry scripts and external references. See what ScanBeforeOpen finds in the source, what preview means here and where the limits are.
Check files for prompt-injection patterns before AI upload
Review known prompt-injection patterns, invisible characters and confidential content before AI uploads. Heuristics and limited text excerpts explained.
Limitations
ScanBeforeOpen detects known risk patterns and helps with triage. It is not a complete antivirus, not a malware sandbox, and not a 100% guarantee.
FAQ
Green means: no obvious known patterns found. Still check the source and context before opening the file.
Save it. Do not open it. Scan first.
The first safe action is simple: choose a file and scan it locally.
Scan a file now