Malicious PDF Generator ☠️

Generate 67 malicious PDF test files for testing phone-home callbacks, SSRF, XSS, XXE, NTLM credential theft, and data exfiltration in PDF viewers, converters, and web applications. Can be used with Burp Collaborator or Interact.sh

Used for penetration testing, bug bounty hunting, and/or red-teaming etc. I created this tool because I needed a tool to generate a bunch of PDF files with various links. Educational and professional purposes only.

Usage

pip install -r requirements.txt
python3 malicious-pdf.py burp-collaborator-url

Output will be written to the output/ directory as: test1.pdf, test2.pdf, test3.pdf etc.

Options

--output-dir DIR    Directory to save generated PDF files (default: output/)
--no-credit         Do not embed credit/attribution metadata in generated PDFs
--obfuscate LEVEL   Obfuscation level (0-3):
                      0 = None (default)
                      1 = PDF name hex encoding + string octal/hex encoding
                      2 = Level 1 + JS bracket notation + javascript: URI case/whitespace obfuscation
                      3 = Level 2 + FlateDecode stream compression

Example with obfuscation:

python3 malicious-pdf.py https://your-interact-sh-url --obfuscate 2

Maximum obfuscation (Level 4 wraps JS payloads in a base64 decoder stub so the original API calls never appear as literal substrings):

python3 malicious-pdf.py https://your-interact-sh-url --obfuscate 4

Purpose

• Test web pages/services accepting PDF files
• Test security products
• Test PDF readers
• Test PDF converters
• Test server-side PDF processing libraries (PDFBox, iText, etc.)
• Test PDF static analysis tools — staged JS payloads (form-field /V, base64 decoder) defeat naïve /JS regex scanners
• Bug bounty hunting — useful for finding SSRF, XXE, blind callbacks, and NTLM leaks in file upload endpoints, PDF-to-image converters, and document processing pipelines on programs that accept PDF input

Credits

• Insecure features in PDFs
• Burp Suite UploadScanner
• Bad-Pdf
• A Curious Exploration of Malicious PDF Documents
• "Portable Document Flaws 101" talk at Black Hat USA 2020
• Adobe Reader - PDF callback via XSLT stylesheet in XFA
• Foxit PDF Reader PoC, DoHyun Lee
• Eicar test file by Stas Yakobov
• Multiple PDF Vulnerabilities - FormCalc & XXE
• PDF - Mess with the web - FormCalc header injection
• Adobe Reader PDF - Client Side Request Injection
• ImageMagick - Shell injection via PDF password
• Portable Data Exfiltration - PortSwigger Research
• CVE-2024-4367 - Arbitrary JS execution in PDF.js
• PDF File Formats Security - Philippe Lagadec
• CVE-2016-2175 - Apache PDFBox XXE
• CVE-2017-9096 - iText XXE
• CVE-2020-29075 - Acrobat Reader silent DNS tracking
• CVE-2022-28244 - Acrobat Reader CSP bypass
• CVE-2018-5158 - Firefox PDF.js PostScript calculator injection
• CVE-2018-20065 - PDFium URI action without user gesture
• ExpMon - Sophisticated Adobe Reader 0-day analysis (April 2026) — inspiration for test33_13/14/15 and obfuscation Level 4

In Media

• Brisk Infosec
• Daily REDTeam
• Malicious PDF File | Red Team | Penetration Testing
• John Hammond - Can a PDF File be Malware?
• Black Hat Ethical Hacking
• 0x1 Pentesting Collection
• Security Toolkit / WADComs
• unsafe.sh
• Cristi Zot on LinkedIn
• Siva R. on LinkedIn

Complete Test Matrix

Todo: New test cases

• Acrobat JS fingerprinting APIs — Add test cases for reconnaissance/fingerprinting APIs used in the April 2026 Adobe 0-day exploit chain (ref): Collab.isDocReadOnly (filesystem probing), app.plugIns (enumerate installed plugins), app.viewerVersion (version fingerprinting)

Todo: Obfuscation methods not yet implemented

• Empty-password PDF encryption — Encrypt all strings/streams with empty user password. Document opens without prompting but static analysis tools cannot read content. Biggest gap in current obfuscation. Ref: Didier Stevens, How secure is PDF encryption?
• Object streams (ObjStm) — Hide PDF objects inside compressed stream containers. Simple parsers (including PDFiD without -O flag) miss objects entirely. Ref: PDF spec ISO 32000 §7.5.7
• getAnnots() code storage — Split JavaScript payload across annotation metadata fields (subject, author). Retrieve at runtime via app.doc.getAnnots()[n].subject and eval. Ref: Julia Wolf - PDF Obfuscation using getAnnots()
• Info dict data extraction — Store encoded payload in /Info trailer fields (/Title, /Author). Retrieve at runtime via info.Title in JS. Ref: corkami PDF tricks
• AcroForm field value extraction — Store payload fragments in form field /V values. Retrieve via getField("name").value in JS. Ref: corkami PDF tricks
• Names tree split execution — Split JavaScript across multiple /Names entries executed sequentially. Ref: corkami PDF tricks
• Incremental updates after %%EOF — Append new objects/actions after the original %%EOF marker via incremental update. Ref: PDF101 content masking, Didier Stevens
• JS unescape() encoding — Wrap JS payload in eval(unescape("%61%6C%65%72%74...")). Ref: corkami PDF tricks
• Fake file headers — Prepend JPEG/HTML/other magic bytes before %PDF- header (spec allows header within first 1024 bytes). Confuses file-type detection. Ref: corkami, Decalage
• Anti-emulation checks — Detect real Adobe Reader via event.target.zoomType == "FitPage" or global variable type checks before executing payload. Ref: corkami PDF tricks

Won't implement

• CVE-2023-26369 - Adobe Acrobat TTF font heap OOB write — Requires binary exploitation (heap spray, ROP chains, shellcode). No public PoC. Cannot produce a simple callback.
• CVE-2021-28550 - Adobe Acrobat Use-After-Free — Requires binary exploitation chain + sandbox escape (CVE-2021-31199/31201). No public PoC. Cannot produce a simple callback.

Star History