Home/Legal Center/Security & Vulnerability Disclosure
Security & Trust

Security & Vulnerability Disclosure

Details our client-side cryptography, in-memory isolation perimeters, and coordinated vulnerability disclosure guidelines.

Effective Date: November 1, 2026•Version 2026.11•Audience: International (Multinational)
Core International Commitment & Document Ownership Guarantee

DocuNexa is an international, browser-based document productivity platform designed for users worldwide. You retain 100% exclusive copyright and ownership of all documents loaded into DocuNexa. Our client-side architecture executes file processing in your personal browser memory. We never own, inspect, sell, or store your documents on our servers.

1. Technical Security Architecture Overview

DocuNexa replaces risky, server-dependent cloud upload architectures with modern client-side execution. By compiling battle-tested open-source engines into WebAssembly (Wasm) and executing tools in browser threads, we eliminate the primary vulnerability of online document processors: the centralized document honeypot.

2. In-Memory Volatile Processing & Data Movement Taxonomy

When a document is opened in DocuNexa, file bytes are read directly into volatile device RAM using standard Web APIs (FileReader and Uint8Array buffers).

  • No temporary file scratchpads or caches are created on remote servers.
  • Binary magic bytes (%PDF-) are checked prior to parsing to block disguised executables.
  • Generated filenames are sanitized against path traversal characters (/, \, :, *, ?, <, >, |).
  • Allocated memory is reclaimed immediately upon closing the browser tab.

Document Isolation & Data Flow Classification

Category A: Pure Local Tools (33 of 34)

PDF Unit Cutter, Merge, Split, Compress, Encrypt, Decrypt, Forms, Conversions, etc. Zero outbound network traffic. 100% in-browser RAM.

Category B: Local WebAssembly + Static Model (OCR PDF)

Tesseract.js WebAssembly downloads static open-source dictionary weights from jsDelivr CDN on first run. Zero document bytes or images leave your browser.

Category C: Local Heuristic Analysis (Summarizer & Translate)

Deterministic structural heuristics & local bilingual glossary. Zero text transmitted to third-party LLMs or cloud AI providers (OpenAI, Anthropic, Google).

Category D: Server File Storage (Zero Tools)

Zero S3 or cloud storage buckets. No server endpoint receives, inspects, or logs document file streams.

3. Cryptographic Standards: PDF AES Encryption vs. Decryption

Precise Cryptographic Scope & Algorithm Distinctions:

DocuNexa does not encrypt all uploaded or processed PDFs with AES-256. When using standard editing tools (such as PDF Unit Cutter, Merge, Split, Rotate, or Compress), document bytes are processed in their normal unencrypted format within temporary typed array buffers (Uint8Array) in local device RAM and released when you close the tab.

Furthermore, DocuNexa distinguishes PDF-standard ISO 32000 AES encryption from Web Crypto AES-GCM: DocuNexa does not use Web Crypto AES-GCM for document processing, as AES-GCM is an envelope authenticated cipher that is incompatible with standard PDF readers (Adobe Acrobat, Apple Preview, Google Chrome). Cryptographic operations are applied strictly and accurately according to their actual implementation:

  • Protect PDF (ISO 32000-1 AES-256 Encryption): Applied exclusively when you choose the Protect PDF tool. DocuNexa invokes the @pdfsmaller/pdf-encrypt WebAssembly engine to compile standard ISO 32000 encryption dictionaries using AES-256 in Cipher Block Chaining (CBC) mode with SHA-256 / MD5 key derivation. Configurable user passwords (required to open) and owner permissions (restricting printing, extraction, and modification) are embedded client-side. The resulting document requires your password in all standard readers.
  • Unlock PDF (Client-Side Decryption & Protection Removal): When opening a password-protected PDF, DocuNexa verifies supplied authorization credentials and removes encryption dictionaries using @pdfsmaller/pdf-decrypt in browser RAM to output an unencrypted PDF. This is local decryption, not encryption. Passwords and decrypted file streams are never transmitted over the network.
  • Server Hashing & Token Cryptography: For non-document security functions (API rate-limiting, Bybit HMAC query signing, Telegram webhook validation, and daily analytics salt rotation), DocuNexa uses standard SHA-256 / HMAC-SHA256 digests.

4. Vulnerability Disclosure Safe Harbor

Safe Harbor for Ethical Security Researchers

DocuNexa values the contributions of the global cybersecurity research community. If you conduct vulnerability research in good faith and in compliance with this policy, we consider your activities authorized. We will not pursue civil lawsuits or initiate criminal complaints against ethical researchers who respect these guidelines.

5. In-Scope & Out-of-Scope Targets

In-Scope Targets
  • Production web application: https://docunexa.pro.et
  • Authentication & Profile APIs (/api/auth/*)
  • Payment & Order verification logic (/api/payments/*)
  • Admin security controls (/api/admin/*)
  • Client-side PDF sandbox escapes or DOM injection
Out-of-Scope
  • Third-party networks: Vercel CDN, Supabase cloud, Bybit exchange
  • Distributed Denial of Service (DDoS/DoS)
  • Social engineering, phishing, or physical attacks
  • Third-party advertising scripts (Monetag)

6. Prohibited Security Testing Methods

  • Never access, modify, or destroy another user's data or account.
  • Never exploit a vulnerability beyond the minimal proof-of-concept necessary to demonstrate its presence.
  • Never execute automated high-volume fuzzers or brute-force scripts against production authentication endpoints.
  • Never disclose vulnerability details publicly before a fix has been verified and deployed by our team.

7. How to Submit a Vulnerability Report

Submit reports by emailing docunexa.security@gmail.com with the subject: [Security Vulnerability Report] <Title>. Include:

1. Vulnerability description and impact assessment (e.g. CSRF, IDOR, Auth Bypass, XSS)

2. Affected endpoint or tool URL

3. Step-by-step reproduction instructions

4. Minimal non-destructive proof-of-concept (PoC) code or HTTP request dump

5. Recommended mitigation or remediation steps

8. Target Response & Remediation Timelines (Non-SLA Estimates)

The following timeframes represent operational target estimates on a best-efforts basis, not guaranteed commercial Service Level Agreements (SLAs):

  • Target Initial Acknowledgment: Aimed within 24 to 48 business hours of receipt.
  • Target Triage & Severity Assessment: Typically conducted within 2 to 3 business days.
  • Target High/Critical Remediation: Prioritized for fix deployment within 7 to 14 business days.
  • Target Medium/Low Remediation: Scheduled in routine maintenance releases within 30 business days.

9. User Account Security Responsibilities

Platform security is a shared responsibility. We strongly encourage all users to:

  • Use unique passwords for your DocuNexa account (never reuse passwords from other services).
  • Ensure your registered email account is secured with multi-factor authentication (2FA).
  • Never share your credentials with third parties or click unverified links.

10. Security Operations Contact

Security Operations: docunexa.security@gmail.com

PGP / Encryption: Inquiries requiring PGP encryption may request our public key via email.

Legal, Privacy & Compliance Contact

For formal inquiries regarding these terms, data protection requests (GDPR/CCPA/UK GDPR), copyright notices, security disclosures, or enterprise agreements, please contact:

Email: docunexa.security@gmail.com

Project: DocuNexa (Independent Software Project)

Operator: DocuNexa Project Maintainers & Contributors

Website: https://docunexa.pro.et

Security & Vulnerability Disclosure — DocuNexa | DocuNexa