naru-config 0.6.4

A security-first configuration manager with encryption and audit logging
# ๐Ÿ›ก๏ธ Security Model

**Version:** 0.6.0  
**Last Updated:** 2026-03-15  

Naru is designed with a **Zero-Trust** philosophy. This document outlines the cryptographic standards and safety measures implemented to protect your sensitive data.

## ๐Ÿ”‘ Cryptography

### 1. Encryption Algorithm
Naru uses **AES-256-GCM** (Advanced Encryption Standard with Galois/Counter Mode).
- **Confidentiality**: Data is encrypted using a 256-bit key.
- **Integrity**: GCM provides an authentication tag that ensures the ciphertext has not been tampered with.
- **Randomness**: Every encryption operation generates a unique 12-byte **Nonce** (number used once) to prevent pattern recognition.

### 2. Key Derivation (KDF)
To ensure strong keys regardless of user password length, Naru employs a **SHA-256 KDF**:
- The `NARU_ENCRYPTION_KEY` environment variable is hashed using SHA-256.
- The resulting 32-byte (256-bit) digest is used as the master encryption key.
- This prevents "short key" vulnerabilities.

## ๐Ÿ›ก๏ธ Input Sanitization

### 1. Directory Traversal Protection
Naru implements a multi-stage path sanitization logic:
- **Null Byte Check**: Rejects any path containing `\0`.
- **Component Analysis**: Every path is broken into components. Any component containing `..` (Parent Directory) is strictly rejected.
- **Absolute Path Rejection**: Naru only operates within relative paths to prevent access to system files (e.g., `/etc/passwd`).
- **Separator Unification**: Handles both `/` and `\` to ensure security across Linux, macOS, and Windows.

### 2. Environment Injection
Environment names and configuration keys are validated against strict alphanumeric patterns:
- **Allowed**: `[a-zA-Z0-0_.-]`
- **Rejected**: Any character that could be used for shell injection (`;`, `&`, `|`, `>`, etc.).

## ๐Ÿ•ต๏ธ Audit Log Integrity
The audit system uses **Hash Chaining**:
- Each log entry includes `previous_hash` and its own `hash`.
- `hash = SHA256(timestamp + action + env + key + values + previous_hash)`.
- Changing a single bit in the past logs invalidates the entire subsequent chain.

## ๐Ÿ›‘ Threat Model & Mitigations

| Threat | Mitigation |
| :--- | :--- |
| **Physical Access** | Config files are encrypted at rest. Audit logs are tamper-evident. |
| **Process Spoofing** | Advisory file locking prevents concurrent process manipulation. |
| **Dictionary Attacks** | Mitigated by SHA-256 KDF (though strong passwords are recommended). |
| **Improper Validation** | Schema enforcement prevents invalid/malicious data types from being stored. |
| **Data Corruption** | Atomic writes and checksums (via GCM tags) detect and prevent corruption. |
| **Race Conditions** | Atomic update API with file locking prevents concurrent write conflicts. |
| **Path Traversal** | Multi-stage sanitization rejects `..`, null bytes, and absolute paths. |
| **Null Byte Injection** | All input validated for null bytes before file operations. |
| **Unicode Attacks** | NFC normalization prevents homograph and encoding bypass attacks. |

## ๐Ÿงช Security Testing

Naru includes a comprehensive security testing suite:

### Penetration Tests
1. **Race Condition Attack** - Tests concurrent write vulnerabilities
2. **Path Traversal Attack** - Validates directory escape prevention
3. **Null Byte Injection** - Tests null byte handling in inputs
4. **Regex DoS (ReDoS)** - Validates regex performance under attack
5. **Audit Log Injection** - Tests log integrity preservation
6. **Secret Masking Bypass** - Validates sensitive data protection
7. **Integer Overflow** - Tests numeric boundary validation
8. **Unicode Normalization** - Validates Unicode handling consistency

### Deep Security Analysis
- **Cryptographic Analysis** - Key zeroization, timing attacks, weak key detection
- **Race Condition Testing** - Old vs new API comparison, lock timeout behavior
- **Information Leak Testing** - Error messages, memory dumps, audit log masking
- **Logic Bug Detection** - Validation bypass attempts, Unicode attacks
- **DoS Testing** - Regex backtracking, memory exhaustion, deep JSON nesting

All 257 security tests pass successfully.