Isolate the carrier
Comments, revisions, alternate bodies, calendar descriptions, hidden sheets, formulas, metadata, and QR payloads become attributable evidence.
Shipped product surface
Scan what the model may parse, not just what a user can read.
Bring a link, file, or selected text in through the Share Sheet. Veridicus Scan inspects the visible content and supported hidden channels, then reports the evidence, risk, guidance, and any partial-coverage boundary it encountered.
Coverage map
The app covers source surfaces, hidden channels, and decision outputs.
The useful question is not “does it scan content?” but “which content, which hidden channels, and what comes back when the scan is done?”
Content surfaces
- Text and web: TXT/plain, HTML/HTM, SVG, MD/Markdown
- Documents and data: PDF, DOCX, XLSX, JSON, CSV
- Mail and calendar: EML and ICS
- Links: URL/WEBLOC files and user-triggered public HTTPS URLs
- Images: PNG, JPG/JPEG, WebP, and GIF
Hidden channels
- DOCX comments, revisions, vanished text, and document metadata
- EML alternative bodies and supported textual attachments
- Calendar descriptions plus XLSX hidden sheets and suspicious formulas
- Webpage comments, hidden content, and metadata
- QR payloads in supported images; animated images inspect the first frame
Detection and outputs
- Base64, hexadecimal, URL-percent, and ROT13 decoding with rescans
- Correlation of suspicious instructions split across artifacts
- Deterministic rules plus an on-device semantic layer, including selected Spanish, Chinese, and Arabic patterns
- Risk, locations, evidence, guidance, and explicit coverage notes for review
- JSON or PDF export with evidence snippets redacted by default
Detection layers
Extraction, decoding, and detection stay separate enough to review.
Veridicus Scan isolates supported hidden channels before evaluation, decodes bounded payloads and rescans them, then combines deterministic checks with an on-device semantic classifier. The result is a reviewed decision surface, not a guarantee beyond the inspected evidence and stated coverage.
Decode and correlate
Base64, hex, URL-percent, and ROT13 content is decoded within bounds, rescanned, and correlated when an instruction is split across artifacts.
Keep rules verifiable
Rule updates are checked manually from Settings, then verified with SHA-256 integrity and an Ed25519 signature before use. Scan content is not sent with an update check.
Coverage discipline
Coverage stays explicit when the scan is bounded.
Partial results, redirect limits, and evidence-redaction defaults stay visible in the result so a user can judge what the scan did and did not prove.
The non-image import window stays visible
Non-image imports use a bounded 20 MB scan window. Larger inputs can produce partial coverage instead of a stronger claim than the inspected bytes support; image imports use their separate image path.
Image limits are stated precisely
Supported images can be inspected for visual signals and QR payloads without the non-image import cap. Animated images inspect the first frame and report that limitation.
Evidence stays usable after export
JSON and PDF output keep findings, guidance, and coverage notes together, with evidence snippets redacted by default.
Related pages
Read the adjacent product surfaces by source, trust model, or output.
Coverage sits in the middle of the product story. The surrounding pages explain why it matters, how the URL path behaves, and how the report communicates the result.
Why it matters
Start with the gap between visible content and parser-visible content before diving into the exact scan surface.
URL scanning
See how strict mode, lenient mode, redirect boundaries, and URL-specific coverage details work.
Reports and exports
See how the scan turns into a readable report with findings, guidance, and PDF or JSON output.
Next step
Take the coverage map into real scenarios.
See how these scan surfaces and outputs show up in assistant safety, document review, team workflows, and local MCP automation.