n8n-nodes-trusera

n8n community node package that scans your workflows for AI security risks using Trusera AI-BOM.

Drop a single node, activate, and visit /webhook/trusera to see a full interactive security dashboard of every AI component in your n8n instance.

Installation

In your n8n instance:

1. Go to Settings > Community Nodes
2. Enter n8n-nodes-trusera
3. Click Install

Or install manually via CLI:

cd ~/.n8n/nodes
npm install n8n-nodes-trusera
# Restart n8n

Quick Start (1 minute)

The Trusera Webhook node gives you a full security dashboard with zero configuration:

1. Create a new workflow
2. Add the Trusera Webhook node
3. Add your n8n API credential (Settings > n8n API > Create API Key)
4. Activate the workflow
5. Visit http://your-n8n-url/webhook/trusera

That's it. One node, full dashboard.

Nodes

This package includes 5 nodes for different use cases:

Trusera Webhook (Recommended)

The one-node solution. This is what most users need.

| | | |---|---| | Type | Trigger (webhook) | | Credential | n8n API (required) | | URL | /webhook/trusera | | Method | GET |

Self-contained trigger node that:

• Fetches all workflows from your n8n instance via the REST API
• Scans every workflow for AI components and security risks
• Serves an interactive HTML dashboard directly at /webhook/trusera

Parameters:

| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | Dashboard Password | string | (empty) | Optional. If set, the dashboard is AES-256-GCM encrypted and visitors must enter this password to view it. |

Setup:

[Trusera Webhook] → (no other nodes needed)

Trusera Dashboard

Use this if you already have a built-in Webhook node configured and want to chain it.

| | | |---|---| | Type | Action | | Credential | n8n API (required) | | Input | Any trigger | | Output | { html, headers, statusCode, body } |

Fetches all workflows, scans them, and returns an HTML dashboard as output. Designed to connect after a built-in n8n Webhook node for custom webhook setups.

Parameters:

| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | Dashboard Password | string | (empty) | Optional AES-256-GCM encryption password. |

Setup:

[Webhook (GET /dashboard, responseMode: lastNode)] → [Trusera Dashboard]

Configure the Webhook node with:

• Response Mode: Last Node
• Response Content Type: text/html
• Response Data: First Entry JSON
• Response Property Name: html

Trusera Scan

Use this for programmatic scanning — pipe workflow JSON in, get structured results out.

| | | |---|---| | Type | Action | | Credential | n8n API (optional) | | Input | Workflow JSON | | Output | Components array with risk scores |

Scans workflow JSON for AI components and returns structured scan results. Use this when you want to process scan results programmatically (e.g., send Slack alerts, store in a database, trigger CI/CD gates).

Operations:

| Operation | Description | |-----------|-------------| | Scan Workflow JSON | Scan a single workflow JSON from input | | Scan Multiple Workflows | Scan an array of workflow JSONs |

Parameters:

| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | Workflow JSON Field | string | json | Input field containing the workflow JSON | | Workflows Array Field | string | workflows | Input field containing the workflows array | | File Path | string | workflow.json | Identifier for the workflow in results |

Setup:

[HTTP Request (GET /api/v1/workflows)] → [Trusera Scan] → [Slack / Email / DB]

Output example:

{
  "components": [
    {
      "name": "GPT-4o Agent",
      "type": "llm_provider",
      "provider": "OpenAI",
      "modelName": "gpt-4o",
      "risk": { "score": 45, "severity": "medium", "factors": [...] },
      "flags": ["no_error_handling", "unpinned_model"]
    }
  ],
  "totalComponents": 1
}

Trusera Policy

Use this to enforce security gates — block deployments with critical findings.

| | | |---|---| | Type | Action | | Input | Scan results from Trusera Scan | | Output | { passed: boolean, violations: string[] } |

Evaluates scan results against configurable security policies. Chain after Trusera Scan to implement CI/CD security gates.

Parameters:

| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | Scan Result Field | string | (empty) | Field containing scan result (empty = entire input) | | Max Critical | number | 0 | Maximum critical-severity components allowed | | Max High | number | -1 | Maximum high-severity components (-1 = unlimited) | | Max Risk Score | number | -1 | Maximum risk score for any component (-1 = unlimited) | | Block Providers | string | (empty) | Comma-separated provider blocklist (e.g., OpenAI,Anthropic) | | Block Flags | string | (empty) | Comma-separated flag blocklist (e.g., hardcoded_api_key,no_auth) |

Setup:

[Trusera Scan] → [Trusera Policy] → [IF passed] → [Deploy] / [Alert]

Output example:

{
  "passed": false,
  "violations": [
    "Found 2 critical components (max: 0)",
    "Blocked provider: OpenAI"
  ],
  "totalComponents": 15
}

Trusera Report

Use this to generate human-readable reports for Slack, email, or documentation.

| | | |---|---| | Type | Action | | Input | Scan results from Trusera Scan | | Output | Markdown or JSON report |

Generates formatted security reports from scan results.

Parameters:

| Parameter | Type | Default | Description | |-----------|------|---------|-------------| | Scan Result Field | string | (empty) | Field containing scan result (empty = entire input) | | Format | options | markdown | markdown or jsonSummary | | Include Low Severity | boolean | false | Whether to include low-severity findings |

Setup:

[Trusera Scan] → [Trusera Report] → [Send Email / Slack Message]

Credentials

n8n API

The Trusera nodes connect to your n8n instance's own REST API to fetch workflows.

| Field | Description | |-------|-------------| | API Key | Your n8n API key (Settings > n8n API > Create API Key) | | n8n Base URL | URL of your n8n instance (default: http://localhost:5678) |

What It Detects

The scanner identifies AI components across your workflows:

• LLM Providers — OpenAI, Anthropic, Google Gemini, Mistral, Groq, Ollama, Azure OpenAI, AWS Bedrock, Cohere, HuggingFace
• Agent Frameworks — n8n AI agents, LangChain chains, ReAct agents
• Tools — Code execution nodes, HTTP request tools, vector stores (Pinecone, Qdrant, Weaviate, ChromaDB, Supabase)
• MCP Clients — Model Context Protocol connections to external servers
• Models — Embedding models, chat models, completion models
• Memory — Buffer memory, conversation memory, session-based memory

Risk Flags

| Flag | Weight | Description | |------|--------|-------------| | hardcoded_api_key | 30 | Hardcoded API key in workflow JSON | | hardcoded_credentials | 30 | Hardcoded credentials in node parameters | | code_http_tools | 30 | Agent with both code execution and HTTP request tools | | shadow_ai | 25 | AI dependency not declared in project files | | webhook_no_auth | 25 | Webhook trigger without authentication | | internet_facing | 20 | AI components exposed to internet via webhook | | multi_agent_no_trust | 20 | Multi-agent system without trust boundaries | | agent_chain_no_validation | 20 | Agent-to-agent chain without output validation | | mcp_unknown_server | 20 | MCP client connected to unknown/external server | | no_auth | 15 | AI endpoint without authentication | | no_rate_limit | 10 | No rate limiting on AI endpoint | | deprecated_model | 10 | Using a deprecated AI model version | | no_error_handling | 10 | No error handling configured for AI calls | | unpinned_model | 5 | Model version not pinned to specific release |

Severity Thresholds

| Severity | Score Range | Color | |----------|-------------|-------| | Critical | 76 - 100 | Red | | High | 51 - 75 | Orange | | Medium | 26 - 50 | Blue | | Low | 0 - 25 | Green |

Dashboard Features

The interactive HTML dashboard includes:

• Summary cards — total components, workflows scanned, highest risk score, scan duration
• Severity distribution chart — donut chart showing critical/high/medium/low breakdown
• Component types chart — bar chart of LLM providers, agents, tools, models, MCP clients
• OWASP LLM Top 10 chart — risk flags mapped to OWASP categories
• Scanned workflows table — each workflow with trigger type, AI component count, highest risk, severity badge
• Findings table — all detected components with name, type, provider, severity, risk score, workflow
• Filters — search by name, filter by severity/type/workflow
• Export — CSV and JSON export buttons
• Dark/light mode — toggle in the header
• Password protection — optional AES-256-GCM encryption

Example Workflows

Minimal: Security Dashboard

[Trusera Webhook] → activate → visit /webhook/trusera

CI/CD Gate: Block Risky Deployments

[Schedule Trigger] → [HTTP Request: GET /api/v1/workflows]
                    → [Trusera Scan]
                    → [Trusera Policy (maxCritical: 0)]
                    → [IF: passed = true]
                        → Yes: [Deploy]
                        → No: [Slack Alert]

Weekly Report: Email Summary

[Schedule Trigger (weekly)] → [HTTP Request: GET /api/v1/workflows]
                             → [Trusera Scan (scanMultiple)]
                             → [Trusera Report (markdown)]
                             → [Send Email]

License

MIT