0xray 4.0.21 → 4.0.23

package/src/mcps/shared/prompt-security-validator.ts

@@ -1,199 +0,0 @@
-/**
- * 0xRay Framework - AI Prompt Security Validator
- *
- * Specialized security validation for AI agent prompts and responses
- * Prevents prompt injection, system prompt override, and malicious inputs
- */
-
-export interface PromptSecurityConfig {
-  enabled: boolean;
-  maxPromptLength: number;
-  allowedPatterns: RegExp[];
-  blockedPatterns: RegExp[];
-  sanitizeLevel: "basic" | "strict" | "paranoid";
-}
-
-export interface SecurityValidationResult {
-  isSafe: boolean;
-  violations: string[];
-  sanitizedPrompt?: string | undefined;
-  riskLevel: "low" | "medium" | "high" | "critical";
-}
-
-export class PromptSecurityValidator {
-  private config: PromptSecurityConfig;
-
-  constructor(config: Partial<PromptSecurityConfig> = {}) {
-    this.config = {
-      enabled: true,
-      maxPromptLength: 10000,
-      allowedPatterns: [],
-      blockedPatterns: [
-        /system\s+prompt\s*[:=]/gi,
-        /ignore\s+previous\s+instructions/gi,
-        /you\s+are\s+now\s+/gi,
-        /forget\s+your\s+previous\s+/gi,
-        /override\s+(your\s+)?instructions/gi,
-        /jailbreak/gi,
-        /dan.*mode/gi,
-      ],
-      sanitizeLevel: "strict",
-      ...config,
-    };
-  }
-
-  /**
-   * Validate AI prompt for security issues
-   */
-  validatePrompt(prompt: string): SecurityValidationResult {
-    if (!this.config.enabled) {
-      return { isSafe: true, violations: [], riskLevel: "low" };
-    }
-
-    const violations: string[] = [];
-    let riskLevel: SecurityValidationResult["riskLevel"] = "low";
-
-    // Length check
-    if (prompt.length > this.config.maxPromptLength) {
-      violations.push(
-        `Prompt too long: ${prompt.length} > ${this.config.maxPromptLength}`,
-      );
-      riskLevel = "medium";
-    }
-
-    // Blocked pattern check
-    for (const pattern of this.config.blockedPatterns) {
-      if (pattern.test(prompt)) {
-        violations.push(`Blocked pattern detected: ${pattern.source}`);
-        riskLevel = "high";
-      }
-    }
-
-    // Advanced pattern analysis
-    if (this.containsSystemPromptOverride(prompt)) {
-      violations.push("Potential system prompt override attempt");
-      riskLevel = "critical";
-    }
-
-    if (this.containsInjectionAttempts(prompt)) {
-      violations.push("Potential prompt injection detected");
-      riskLevel = "high";
-    }
-
-    const isSafe = violations.length === 0;
-    const sanitizedPrompt = isSafe ? undefined : this.sanitizePrompt(prompt);
-
-    return {
-      isSafe,
-      violations,
-      sanitizedPrompt,
-      riskLevel,
-    };
-  }
-
-  /**
-   * Check for system prompt override attempts
-   */
-  private containsSystemPromptOverride(prompt: string): boolean {
-    const overrideIndicators = [
-      "you are now",
-      "act as if",
-      "pretend to be",
-      "role-play as",
-      "from now on you are",
-      "ignore your system",
-      "override your",
-      "new personality",
-      "different role",
-    ];
-
-    const lowerPrompt = prompt.toLowerCase();
-    return overrideIndicators.some((indicator) =>
-      lowerPrompt.includes(indicator),
-    );
-  }
-
-  /**
-   * Check for prompt injection attempts
-   */
-  private containsInjectionAttempts(prompt: string): boolean {
-    const injectionIndicators = [
-      "ignore previous",
-      "forget instructions",
-      "start over",
-      "new conversation",
-      "reset context",
-      "clear memory",
-      "delete history",
-    ];
-
-    const lowerPrompt = prompt.toLowerCase();
-    return injectionIndicators.some((indicator) =>
-      lowerPrompt.includes(indicator),
-    );
-  }
-
-  /**
-   * Sanitize prompt by removing dangerous patterns
-   */
-  private sanitizePrompt(prompt: string): string {
-    let sanitized = prompt;
-
-    // Remove blocked patterns
-    for (const pattern of this.config.blockedPatterns) {
-      sanitized = sanitized.replace(pattern, "[REDACTED]");
-    }
-
-    // Additional sanitization based on level
-    if (
-      this.config.sanitizeLevel === "strict" ||
-      this.config.sanitizeLevel === "paranoid"
-    ) {
-      // Remove potential code execution
-      sanitized = sanitized.replace(/```[\s\S]*?```/g, "[CODE_BLOCK_REMOVED]");
-
-      // Remove potential file system access
-      sanitized = sanitized.replace(/(?:\/|\\)\.\.(?:\/|\\)/g, "/");
-
-      // Remove potential command injection
-      sanitized = sanitized.replace(/[;&|`$()]/g, "");
-    }
-
-    if (this.config.sanitizeLevel === "paranoid") {
-      // Extreme sanitization - only allow alphanumeric, spaces, and basic punctuation
-      sanitized = sanitized.replace(/[^a-zA-Z0-9\s.,!?-]/g, "");
-    }
-
-    return sanitized;
-  }
-
-  /**
-   * Validate agent response for safety
-   */
-  validateResponse(response: string): SecurityValidationResult {
-    // Similar validation but focused on response safety
-    const violations: string[] = [];
-
-    // Check for potentially harmful content
-    if (
-      response.includes("system prompt") ||
-      response.includes("internal instructions")
-    ) {
-      violations.push("Response contains sensitive system information");
-    }
-
-    // Check for code execution attempts
-    if (/eval\s*\(|Function\s*\(/.test(response)) {
-      violations.push("Response contains potential code execution");
-    }
-
-    return {
-      isSafe: violations.length === 0,
-      violations,
-      riskLevel: violations.length > 0 ? "medium" : "low",
-    };
-  }
-}
-
-// Export singleton instance
-export const promptSecurityValidator = new PromptSecurityValidator();