Use Claude to map supplier files through n8n
Felix worked with a colleague using Claude Opus to build a retail data workflow through a self-hosted n8n instance.
dev.to
Original prompt
Shared by the source author. See source notes for context.
I need a sub-workflow that takes raw supplier files with 13 columns and maps them into our 47-column BSMS_ITEM_UPLOAD format.
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Felix
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- Felix reports implemented supplier mapping and validation, with humans correcting barcode behavior and resolving an Apps Script race before production sign-off.
- Human involvement
- The colleague resolved domain issues and signed off production.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Supply reference supplier files and target schema.
- Inspect mappings through n8n MCP.
- Repair data errors with human production sign-off.
Felix worked with a colleague using Claude Opus to build a retail data workflow through a self-hosted n8n instance. The business task was turning inconsistent supplier files into records suitable for Oracle EBS. Through MCP, Claude could inspect node configurations and run workflows while also reading reference files and checking Sheets.
The agent examined how the team converted a supplier file with thirteen columns into a forty-seven-column upload format. It identified field mappings, recognized alternate barcode names and helped build season-resolution logic. The larger workflow included file routing, extraction, merging, validation and spreadsheet output.
Debugging exposed both useful analysis and agent mistakes. Claude found cached spreadsheet column names that produced empty rows despite a success status. Its own fuzzy matching also mishandled empty barcodes until the humans described the symptom. An Apps Script race condition required the colleague to introduce a flag and teach the pattern back to Claude. Felix explicitly places the human architect in charge of production sign-off. The result was a implemented workflow with domain corrections, not a fully unattended migration.
Prompt
Original prompt
Source: dev.to · Felix
Frequently asked questions
Which file formats were mapped?
Thirteen-column inputs to forty-seven-column uploads.
What spreadsheet issue was found?
Cached columns caused apparently successful empty writes.
Who resolved the script race?
The human colleague supplied a flag-based fix.
More details
About this account
Claude inspected live configs, mapped fields and revised workflows.
Firsthand account opened and read; outcomes attributed to the author. Prompt is a verbatim excerpt of the shared original, not the complete instruction.
Credit: Felix
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Opus · n8n MCP · Google Drive · Google Sheets · Chrome
Supplier data · Oracle EBS · n8n MCP
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.