This guide walks you through building a chat-style workflow that lets an n8n AI Agent call Explorium’s Model Context Protocol (MCP) tools, powered by Anthropic’s Claude.
You’ll end up with a flow like this:

1. Prerequisites 🔧
Requirement | Notes |
---|---|
n8n @ v1.31+ | Self-hosted or Cloud |
Explorium API key | Get one from your AgentSource Admin Panel |
Anthropic API key | Needed for Claude 3.x Sonnet |
SSE endpoint | https://mcp.explorium.ai/sse |
2. Create Your Flow
- Trigger – When chat message received (built-in)
- AI Agent – type: Tools Agent
- Anthropic Chat Model – Claude 3.7 Sonnet
- Memory – simple in-flow memory
- MCP Client – calls Explorium tools
Connect the nodes just like the overview image (dotted lines = resource connections).

3. Configure Each Node
3.1 AI Agent configuration
Setting | Value |
---|---|
Agent | Tools Agent |
Source for Prompt | Connected Chat Trigger Node |
System Message | Use a Claude-style system prompt. Example: The assistant is Claude, created by Anthropic. The current date is {{ Date() }}.\nYou are powered by Explorium’s MCP tools. Always call tools when data is required… |
Chat Model | Connect the Anthropic Chat Model node |
Tool | Connect the MCP Client node |
Memory | Connect the Memory node |

3.2 Anthropic Chat Model
- Credential: the Anthropic account you just created
- Model: Claude 3.7 Sonnet
- Enable Thinking: On (optional, shows the “thinking” indicator)

3.3 Anthropic credential
Enter your Anthropic key and test – you should see Connection tested successfully.
3.4 MCP Client node
Field | Value |
---|---|
SSE Endpoint | https://mcp.explorium.ai/sse |
Authentication | Bearer Auth |
Credential for Bearer Auth | Bearer Auth account |
Tools to include | All |

Enter your api key in the Bearer Token field:

4. Test It Out
- Click ▶ Open chat at the bottom of the canvas.
- Ask something data-centric, e.g.
- “Who is the CMO of Snowflake?”
- You should see the MCP Client fire Match Businesses ➜ Enrich Businesses and Claude respond with a neatly-formatted answer: