n8n-nodes-antijection

Antijection is a powerful AI security node for n8n that helps protect your LLM applications from prompt injections, jailbreaks, and safety violations.

Antijection provides real-time analysis and filtering for AI-generated and user-submitted content, ensuring your workflows remain secure and compliant.

Features

• Prompt Injection Detection: Identify and block malicious attempts to manipulate LLM behavior.
• Precision Models:
  • Injection Guard (Fast): Optimized for speed, ideal for English-only real-time applications.
  • Injection Guard Multi: Robust multilingual support for global applications.
  • Safety Guard: Comprehensive analysis for toxic content, PII leaks, and safety risks.
• Customizable Heuristics: Fine-tune protection with enabled/disabled categories and custom keyword blocking.
• Production-Ready: Comprehensive error handling, input validation, and user-friendly messages.

Installation

To install this node in your n8n instance:

1. Go to Settings > Community Nodes.
2. Click Install a new node.
3. Enter n8n-nodes-antijection.
4. Click Install.

Alternatively, for manual installations in your n8n directory:

npm install n8n-nodes-antijection

Operations

Detect

Analyze a text prompt for security risks and safety issues.

Parameters:

• Prompt: The text content you want to analyze (1-10,000 characters).
• Detection Method: Choose between Injection Guard, Injection Guard Multi, or Safety Guard.
• Rule Settings:
  • Enabled: Toggle heuristic rule processing.
  • Disabled Categories: Select specific threat categories to ignore (useful for coding assistants).
  • Blocked Keywords: Add custom strings or regex patterns that should always trigger a block (one per line).

Response Fields:

• risk_score (0-100): Risk level where 0 = safe, 100 = unsafe. Recommended: Block prompts with risk_score ≥ 50.
• detection_method: The method used for detection.
• credits_used: Credits consumed based on token count and method.
• tokens_used: Number of tokens in the prompt.
• categories: List of safety categories detected (SAFETY_GUARD only).
• latency_ms: Response time in milliseconds.

Credentials

To use this node, you need an API Key from the Antijection dashboard.

1. Sign up/Log in at antijection.com.
2. Generate an API Key in your account settings.
3. In n8n, create a new Antijection API credential and enter your key.

Usage Examples

Example 1: Basic Prompt Filtering

1. Trigger (e.g., Webhook)
2. Antijection Node
   - Prompt: {{$json.userInput}}
   - Detection Method: INJECTION_GUARD_MULTI
3. IF Node
   - Condition: {{$json.risk_score}} >= 50
   - True: Send "Blocked" response
   - False: Continue to LLM

Example 2: Coding Assistant (Disable Code-Related Rules)

1. Trigger
2. Antijection Node
   - Prompt: {{$json.query}}
   - Detection Method: INJECTION_GUARD
   - Rule Settings:
     - Disabled Categories: 
       - SQL Injection
       - Command Injection
       - XSS Patterns
       - Path Traversal
3. Continue to code generation LLM

Example 3: Content Moderation with Safety Guard

1. Trigger
2. Antijection Node
   - Prompt: {{$json.userComment}}
   - Detection Method: SAFETY_GUARD
3. IF Node
   - Condition: {{$json.risk_score}} >= 50
   - True: Flag for review ({{$json.categories}})
   - False: Publish content

Available Rule Categories

When using Disabled Categories, you can select from:

• Ignore Instructions - Direct attempts to override system prompts
• System Override - Attempts to toggle developer/admin modes
• Prompt Extraction - Attempts to leak the system prompt
• Role Hijacking - Forcing the AI into a specific persona
• Encoded Attacks - Base64, Hex, or Unicode encoding tricks
• Fuzzy Matches - Common misspellings of attack keywords
• Many Shot - Overloading context with fake Q&A
• SQL Injection - Common SQL injection patterns
• Command Injection - Shell command execution attempts
• XSS Patterns - Script injection and XSS vectors
• Path Traversal - File system traversal attempts
• Unusual Punctuation - Abnormal clusters of special characters
• Repetition Attacks - Excessive or interspersed repetition
• Emojis - Suspicious or excessive use of emojis

Best Practices

Detection Method Selection

• INJECTION_GUARD: Use for high-volume, English-only applications (1 credit/500 tokens)
• INJECTION_GUARD_MULTI: Use for multilingual applications (2 credits/500 tokens)
• SAFETY_GUARD: Use for comprehensive content moderation (4 credits/500 tokens)

Blocked Keywords

Use the Blocked Keywords field to add custom terms specific to your application:

internal_project_name
<special_token>
\b(secret|api_key)\b

Supports Python-style regex patterns for advanced matching.

Troubleshooting

"Authentication failed"

• Cause: Invalid API key
• Solution: Check your Antijection API credentials in n8n. Ensure you've copied the full API key.

"Rate limit exceeded"

• Cause: You've exceeded your API rate limit or credit quota
• Solution: Upgrade your plan at antijection.com or wait before retrying.

"Prompt is too long"

• Cause: Prompt exceeds 10,000 characters
• Solution: Truncate or split your prompt into smaller chunks.

"Invalid request"

• Cause: Malformed request parameters
• Solution: Check that your prompt is not empty and detection method is valid.

Credential Test Fails

• Cause: API key is invalid or network issues
• Solution: Verify your API key is active in the Antijection dashboard. Check network connectivity.

Error Handling

The node provides detailed error messages for common issues:

• 401: Authentication failed - check API key
• 403: Access forbidden - verify permissions
• 429: Rate limit exceeded - upgrade plan or wait
• 400: Invalid request - check input parameters
• 500/502/503: Service temporarily unavailable - retry later

Enable "Continue On Fail" in n8n to handle errors gracefully in your workflow.

Privacy & Terms

Your security and privacy are our top priorities.

• Privacy Policy: https://www.antijection.com/privacy
• Terms of Service: https://www.antijection.com/terms

Resources

• Antijection Website
• API Documentation
• GitHub Repository
• n8n Community Forum

Support

For questions, troubleshooting, or custom enterprise requirements, please contact us at hello@antijection.com.

---

Developed by AITEERA LLC.