Monitor n8n Workflows with Agent Inspector

What You'll Set Up

Agent Inspector acts as a proxy between n8n's LLM nodes and Anthropic/OpenAI APIs. You'll configure n8n to route LLM calls through Agent Inspector, which captures and analyzes every interaction without modifying your workflows.

Step 1: Start Agent Inspector

Launch Agent Inspector with the Anthropic provider (or OpenAI if you're using GPT models):

# For Anthropic
uvx agent-inspector anthropic

# For OpenAI
uvx agent-inspector openai

This starts two services:

• Proxy: http://localhost:4000 - Routes LLM calls
• Dashboard: http://localhost:7100 - View analysis results

Step 2: Configure n8n Credentials

Configure your n8n AI nodes to route through Agent Inspector. This can be done through credential settings for Anthropic/OpenAI nodes, or via HTTP Request nodes for custom setups.

Option A: Using AI Node Credentials (Anthropic/OpenAI)

The easiest way is to update your AI node credentials in n8n:

1. Go to Settings → Credentials in n8n
2. Select your Anthropic API or OpenAI credential
3. Configure the base URL:
   • API Key: Your real Anthropic/OpenAI API key
   • Base URL (Anthropic): http://127.0.0.1:4000
   • Base URL (OpenAI): http://127.0.0.1:4000
4. Click Test to verify the connection
5. Save the credential

n8n Anthropic credential configuration with Agent Inspector proxy

Option B: Using Agent Workflow IDs (Advanced)

To group sessions and link them to static analysis, include a workflow ID in the base URL:

http://127.0.0.1:4000/agent-workflow/{AGENT_WORKFLOW_ID}

Use the same agent_workflow_id when running static scans via MCP tools to correlate findings.

Step 3: Run Your Workflow

Execute any n8n workflow that uses LLM nodes (Anthropic, OpenAI, or other providers configured through Agent Inspector). Here's an example workflow structure:

Example n8n workflow with Anthropic nodes and Code Tools

Step 4: View Analysis in Dashboard

Open http://localhost:7100 in your browser to see real-time analysis:

Sessions List

Every LLM call from n8n appears as a session. You can see status, duration, token usage, and tool calls:

Sessions list showing all captured n8n LLM calls

Security and Behavioral Assessment

Get an at-a-glance view of your agent's security posture and behavioral patterns. This comprehensive assessment includes checks across resource management, behavioral stability, environment configuration, and privacy compliance:

Security assessment showing critical checks, warnings, and behavioral metrics

Environment & Supply Chain

View which models are being used, whether they're pinned to specific versions, and tool usage coverage:

Model version tracking and tool adoption analysis

Privacy & PII Detection

Agent Inspector automatically scans for sensitive data in prompts and responses:

Privacy & PII Compliance report showing detection findings and exposure rates

Behavioral Analysis

After running multiple sessions, Agent Inspector automatically groups them into behavioral clusters and identifies outliers. This helps you understand operational patterns and spot anomalies:

Behavioral clustering showing distinct patterns and outlier detection

What You Get

Common Use Cases

1. Debugging Tool Call Issues

When your LLM isn't calling n8n Code Tools as expected, Agent Inspector shows the exact tool schemas sent to the LLM and which tools were actually invoked.

2. Optimizing Token Usage

Track token consumption across workflow executions to identify prompts that need optimization or tool calls that are inefficient.

3. Testing Workflow Variations

Run the same workflow with different inputs and see how behavioral patterns change. Agent Inspector's clustering helps you understand consistency.

4. Pre-Production Security Review

Before deploying workflows to production, use Agent Inspector to generate security reports showing PII exposure, resource bounds, and audit compliance.

Next Steps