n8n MCP Store Support Copilot w/ OpenRouter & Slack Feedback

Turn n8n into an AI-assisted store support copilot with OpenRouter + Slack feedback

This n8n automation workflow builds a store-support MCP server backed by n8n Data Tables and a chat-based AI copilot powered by OpenRouter. Support teams can look up orders, create returns with policy checks, and automatically capture tool-gap feedback—optionally posting it to Slack.

What this workflow does

• Exposes an MCP tool endpoint with bearer authentication, offering: find_orders, get_order, get_returns, create_return, and send_feedback.
• Fetches orders and returns from n8n Data Tables using the provided customer email or order number.
• Creates returns with validation: when create_return is called, it verifies delivered status, a 30-day window from delivery, excludes final sale items, and checks remaining returnable quantity—then calculates the refund.
• Generates an RMA-style return ID and stores the return request in the returns Data Table, or returns a structured refusal with a policy code and details.
• Captures tool-gap feedback via send_feedback: validates required fields, rate-limits to 30 reports per hour, and scrubs secrets and email addresses before saving to a feedback Data Table.
• Posts feedback to Slack to a configured channel and saves the Slack timestamp back to the feedback record, without failing if Slack is unavailable.
• Runs a chat trigger AI Agent (OpenRouter model) that uses the MCP endpoint as its only tool to answer support questions and file feedback when tool gaps block resolution.

Use cases

• Support agents answer order-status questions by searching with email or order number via MCP tools.
• Automated, policy-safe return creation (including refusal reasons) with RMA ID generation.
• Continuous improvement: collect anonymized tool-gap feedback and route it to Slack for faster tooling updates.

Technical details

• n8n Data Tables: orders, returns, and feedback storage.
• Integrations / nodes used: if, set, slack, crypto, switch, split out.
• Security: bearer-authenticated MCP endpoint.
• AI layer: chat trigger AI Agent using OpenRouter with MCP tools as the sole action interface.