Security Audit Report

Internal security audit findings and remediations for SecureCall · Last updated February 2026

Audit Overview

An internal security audit was conducted on February 15, 2026 covering all three components of the SecureCall platform: the Android client (Kotlin), the Rust cryptographic engine, and the Node.js signaling backend. The audit examined code quality, cryptographic correctness, transport security, authentication mechanisms, and resilience against known attack vectors.

Findings Overview

Remediation Status

Critical Vulnerability Details

The following seven critical vulnerabilities were identified during the audit. All have been remediated and verified through code review and regression testing. Each finding represented a direct threat to the confidentiality or integrity of encrypted communications.

Android Client Remediations

The Android client underwent the most extensive remediation effort, addressing four critical and six high-severity findings. The core issue was that encryption stubs from early development had not been fully replaced with production implementations.

• All crypto stubs replaced with real implementations backed by the Rust native library (libcore_crypto.so)
• Plaintext fallback paths removed entirely — the app will not function if the crypto engine fails to load
• FakeX25519 class replaced with actual X25519 key exchange via the FFI bridge
• WebSocket endpoint now configurable via BuildConfig with WSS (TLS) enforcement for all flavors
• java.util.Random replaced with java.security.SecureRandom in all security-sensitive contexts
• Key zeroization added throughout the codebase — ephemeral keys are cleared from memory after use
• Replay detection enforced with boolean return values (previously void, silently ignoring replays)
• Nonce overflow now throws SecurityException instead of wrapping to zero

Rust Crypto Engine Remediations

The Rust crypto engine had one critical and three high-severity findings, primarily related to memory safety at the FFI boundary and incomplete key zeroization.

• Output buffer capacity validation added before all ptr::copy_nonoverlapping operations in FFI functions
• #[zeroize(skip)] annotation removed from aead_key field — key material is now properly zeroized on drop
• All FFI intermediate values wrapped in Zeroizing<> containers to prevent key material from remaining in memory
• ZeroizeOnDrop derive macro added to IdentityKeyPair for automatic cleanup

Backend Server Remediations

The backend server had two critical and nine high-severity findings. The critical issues were hardcoded credentials and exposed debug endpoints. High-severity findings covered authentication gaps and missing rate limiting.

• Hardcoded TURN credentials removed — server now requires environment variables and fails to start without them
• Debug endpoints (status dumps, connection listings, test routes) removed entirely from the codebase
• Authentication added to Public Key Directory (PKD) and WebSocket signaling paths
• Origin validation implemented to prevent cross-origin WebSocket hijacking
• maxPayload limit set to 64 KB to prevent denial-of-service via oversized messages
• Rate limiting implemented on all public endpoints
• Call participant verification added — only registered users can join call rooms
• Per-IP connection limits enforced to prevent resource exhaustion attacks

Medium-Severity Findings

All 23 medium-severity findings have been documented and filed as GitHub Issues for tracking. These items represent defense-in-depth improvements that do not pose immediate security risks but should be addressed in upcoming development cycles.

Full Cross-Device Emulation Test Results

Comprehensive field test conducted on 3 physical Samsung devices across all 3 subscription tiers. Every feature tested systematically with pass/fail logging.

Test Environment

Phase 1 — Connection Stability

Phase 2 — Call Signaling

Phase 3 — Call Edge Cases

Phase 4 — Contacts & Status

Phase 5 — Feature Gating

Phase 6 — Special Features

Complete Audit Documentation

The complete audit report with detailed vulnerability descriptions, reproduction steps, code references, and verification evidence is available in the project repository:

docs/SECURITY_AUDIT_REPORT.md — Full security audit report with code-level details

For responsible disclosure of new security vulnerabilities, please refer to our Security Policy.