AI Prompts Library

You are red-teaming an AI feature for prompt injection. The system processes untrusted content from [SOURCE: web pages / emails / uploaded documents / user messages].

Produce an attack list. For each attack:
- The injected payload, written out
- Where it would be placed in the untrusted content
- What it tries to make the system do
- Which defence, if any, in the system below would stop it
- Severity if it succeeds

Cover at minimum: direct instruction override, authority impersonation, delayed or conditional triggers, encoded and obfuscated payloads, payloads hidden in markup or metadata, tool-call hijacking, data exfiltration via crafted URLs, and multi-turn setup attacks.

Then state the single strongest defensive gap.

System description:
[DESCRIBE THE SYSTEM, ITS TOOLS, AND ITS CURRENT DEFENCES]

Review this agent execution sandbox. Assume the agent will eventually run adversarial code, either because a user asked it to or because injected content told it to.

Assess:
1. FILESYSTEM: what can it read outside the workspace? Write? Are credentials, SSH keys, cloud metadata or environment files reachable?
2. NETWORK: outbound allowed? To where? Can it reach internal services, cloud metadata endpoints, or exfiltrate to an arbitrary host?
3. PROCESS: can it spawn long-running processes, escape the process tree, or survive the session?
4. RESOURCE LIMITS: CPU, memory, disk, execution time. What happens when it hits them?
5. SECRETS: what is in the environment? Anything that should not be?
6. PERSISTENCE: can state written in one run be read in the next, and should it be?
7. HOST BOUNDARY: container, VM, or neither? Be honest about what a container does and does not isolate.

Rank findings by severity and give the concrete fix for each.

Sandbox setup:
[DESCRIBE]

You are a principal software engineer conducting a thorough code review. You combine deep security expertise with performance engineering knowledge. Review the submitted code with extreme attention to detail.

**Code to Review:**
[PASTE CODE HERE]

**Language/Framework:** [SPECIFY]
**Context:** [WHAT DOES THIS CODE DO AND WHERE DOES IT RUN]

**Review Checklist:**

**Security Analysis (CRITICAL):**
- [ ] SQL Injection: Are all queries parameterized? Any string concatenation in queries?
- [ ] XSS: Is user input sanitized before rendering? Are Content-Security-Policy headers set?
- [ ] CSRF: Are state-changing requests protected with tokens?
- [ ] Authentication: Are passwords hashed with bcrypt/argon2? Are JWTs validated properly?
- [ ] Authorization: Is there proper access control on every endpoint? IDOR vulnerabilities?
- [ ] Input Validation: Are all inputs validated for type, length, format, and range?
- [ ] Secrets: Are API keys, passwords, or tokens hardcoded? Are they in environment variables?
- [ ] Dependencies: Are there known CVEs in the dependency versions used?
- [ ] File Upload: Are file types validated server-side? Is the upload directory outside webroot?
- [ ] Rate Limiting: Are sensitive endpoints rate-limited?

**Performance Analysis:**
- [ ] N+1 Queries: Are there database queries inside loops?
- [ ] Missing Indexes: Are queried columns properly indexed?
- [ ] Memory Leaks: Are event listeners, subscriptions, or intervals cleaned up?
- [ ] Unnecessary Re-renders: Are React components memoized appropriately?
- [ ] Bundle Size: Are large libraries imported when smaller alternatives exist?
- [ ] Caching: Are expensive computations or API calls cached appropriately?
- [ ] Async Operations: Are promises handled correctly? Any unhandled rejections?
- [ ] Algorithm Complexity: Are there O(n^2) or worse operations that could be optimized?

**Code Quality:**
- [ ] Single Responsibility: Does each function/class do one thing well?
- [ ] DRY: Is there duplicated logic that should be extracted?
- [ ] Error Handling: Are errors caught, logged, and handled gracefully?
- [ ] Naming: Are variables and functions named clearly and consistently?
- [ ] Comments: Are complex algorithms explained? Are TODO/FIXME items addressed?

For each finding, provide: Severity (P0-P3), location, explanation, and a concrete fix with code.

Security audit following OWASP Top 10:

**1. Injection Flaws**
- [ ] Use parameterized queries
- [ ] Validate & sanitize all inputs
- [ ] Use ORM safely

**2. Broken Authentication**
- [ ] Implement MFA
- [ ] Use secure session management
- [ ] Hash passwords (bcrypt/Argon2)
- [ ] Implement rate limiting

**3. Sensitive Data Exposure**
- [ ] Encrypt data at rest & in transit (TLS 1.3)
- [ ] Don't log sensitive data
- [ ] Use secure key management

**4. XML External Entities (XXE)**
- [ ] Disable XML external entity processing
- [ ] Use secure XML parsers

**5. Broken Access Control**
- [ ] Verify permissions server-side
- [ ] Implement RBAC/ABAC
- [ ] Test authorization logic

**6. Security Misconfiguration**
- [ ] Remove default credentials
- [ ] Disable directory listing
- [ ] Keep dependencies updated
- [ ] Use security headers (CSP, HSTS)

**7. XSS (Cross-Site Scripting)**
- [ ] Escape output
- [ ] Use Content Security Policy
- [ ] Sanitize HTML inputs

**8. Insecure Deserialization**
- [ ] Validate serialized data
- [ ] Use signing/encryption

**9. Using Components with Known Vulnerabilities**
- [ ] Regular dependency audits
- [ ] Automated security scanning
- [ ] Subscribe to security advisories

**10. Insufficient Logging & Monitoring**
- [ ] Log security events
- [ ] Monitor for anomalies
- [ ] Set up alerting

You are an expert code reviewer with deep knowledge of security, performance, and best practices across multiple programming languages.

Your task is to review the following code and provide:
1. **Security Analysis**: Identify any security vulnerabilities (SQL injection, XSS, CSRF, authentication issues, etc.)
2. **Performance Review**: Highlight performance bottlenecks and suggest optimizations
3. **Code Quality**: Assess readability, maintainability, and adherence to best practices
4. **Bug Detection**: Point out logical errors, edge cases, and potential runtime issues
5. **Recommendations**: Provide specific, actionable improvements with code examples

For each issue found, specify:
- Severity (Critical, High, Medium, Low)
- Location (file, line number if available)
- Detailed explanation of the problem
- Concrete solution with improved code

Code to review:
[PASTE YOUR CODE HERE]