Instruction-like content
Text that appears to direct an AI, change its task or influence how it handles the project.
PDF security for AI projects
Scan PDFs before using them with AI.
A PDF can contain more than the page you see. .mdSiren is designed to check supported visible, hidden and structural content before the file becomes part of an AI project.
Product coming soon- OutsideOutside PDF
- .mdSiren scanReview before use
- After reviewAI project
The PDF problem
The visible page is only one layer.
PDFs are containers for text, layout, links, metadata and other machine-readable structures. Content that is difficult for a person to notice may still be extracted when the file is used with an AI tool.
The useful question is not only “Does this PDF look normal?” It is also “What might the AI read, and should any of it be reviewed first?”
Understand prompt injection in documentsWhat deserves attention
Review the document as a file, not only as a page.
.mdSiren is designed to bring supported document surfaces into one pre-use review.
Hidden and machine-readable layers
Supported content that may be concealed, encoded or otherwise easy to miss during ordinary visual review.
Links and structural features
Supported remote links, active features and PDF conditions that may require a closer look.
A safer PDF intake
Put the review before the upload.
Keep the boundary visible whenever a PDF moves from an outside source into AI-assisted work.
- 1Receive or download the PDF.
- 2Open .mdSiren before adding the file to an AI project.
- 3Run the supported document checks.
- 4Review the status, findings and coverage.
- 5Decide whether the PDF should enter the project.
- 6Scan again if the file changes.
A scan supports a human decision. It does not prove that a document is harmless or replace normal source, privacy and security checks.
Before AI use
Make the intake decision explicit.
A short checklist helps stop “looks fine” from becoming the only document-security control.
- Confirm where the PDF came from and why the project needs it.
- Scan the file before it enters the AI tool or project.
- Read every supported finding in its document context.
- Check which scans completed and where coverage was limited.
- Keep the original unchanged while deciding what to use.
.mdSiren is coming soon. Final product wording and coverage will be checked against the release build before the scanner is linked from this site.
Related resources
Before the project
Give every outside PDF a checkpoint before AI use.
Review what the AI may read, what was checked and what needs your attention before the file enters the project.