Skill Claude · ChatGPT / MCP
Clean product data
Normalize messy product data into a consistent schedule.
What it does
Give Norma a messy FF&E schedule. Norma cleans inconsistent casing, categories, dimensions, language, materials vocabulary and price formatting, then shows you the changes. You choose what to keep before any save that needs your approval.
Install
Connect Arch Studio in Claude or ChatGPT, then return to your conversation and ask Norma for the task. Your assistant uses its available file access and tools to carry out the skill.
Add to Claude
Add to ChatGPTVersions and source
Arch Studio 1.5.1.2 · MCP
Available as MCP workflow instructions in compatible Claude and ChatGPT connections. The assistant performs the work with its own tools and permissions and verifies the actual result before reporting completion.
Arch Studio 1.5.1.2 · Open source
Included in the public plugin for Claude Code and Codex. In this version: Clean product data — Standardize dimensions, units, materials and formatting in an FF&E CSV schedule.
Usage
Ask Norma in plain language. With Arch Studio connected through MCP, no $ or / command is required. Norma is your assistant working with Arch Studio.
Use Arch Studio to clean this sample row: category “chairs”; dimensions “W: 80, D: 60, H: 75 cm”; price “1.234,56 EUR”; material “Madera”. Keep the original values in the review and show a before/after table. Flag ambiguous values and possible duplicates before saving.The full catalog lives in your assistant once Arch Studio is connected. Ask: Norma, what can you do? List the skills, tools, agents and workflows available in this host, grouped by practice area.
What it cleans
Casing
Norma standardizes collection, material and finish casing while preserving brand abbreviations such as HAY and USM.
Categories
Norma maps free text to Arch Studio’s current product categories. For example, “chairs” maps to “Chair” and “mesa” to “Table”. Ambiguous categories are flagged for your review.
Dimensions
Norma separates width, depth, height and units when the source identifies them. The original unit is kept unless you request a conversion. Missing axis labels or units stay unresolved.
| Source | Normalized fields |
|---|---|
| 32 × 24 × 30 in, explicitly labeled W × D × H | W: 32 · D: 24 · H: 30 · Unit: in |
| Ancho: 80, Prof: 60, Alto: 75 cm | W: 80 · D: 60 · H: 75 · Unit: cm |
| W32 D24 H30, no unit supplied | Unit unresolved; raw value retained |
| 80 × 60 × 75 cm, no axis order supplied | Axis assignment unresolved; raw value retained |
Language
Norma translates category, material and finish terms to English unless you choose another language. “Madera” becomes “Wood”; product names and brands stay as written.
Materials and finishes
Norma standardizes supported vocabulary, such as “Stainless” to “Stainless steel”, while preserving the source evidence and selected finish.
Price and currency
Norma separates amounts from currencies and normalizes supported number formats. “1.234,56 EUR” becomes amount 1234.56, currency EUR. A dollar sign alone does not establish the currency.
Duplicates
Norma flags matching names/brands or URLs as possible duplicates. You decide what to keep; a shared product-family URL can refer to different selections.
Whitespace and formatting
Norma removes stray spaces, line breaks and trailing separators so the selected fields use consistent formatting.
Workflow
- Load — Norma reads the schedule and identifies its rows, columns and existing values.
- Analyze — Norma counts the issues and prepares a cleanup preview.
- Review — You review the proposed changes and resolve ambiguous fields or possible duplicates.
- Apply — Norma applies the selected cleanup while preserving unrelated fields and user choices.
- Check — Norma checks the resulting fields and shows what changed.
- Save — Norma delivers the cleaned file or conversation result. For a requested reusable-library update, Norma uses Product Library to save and read back the records.
Pairs with
Product specs from URLs. Norma can extract specifications from product URLs first, then clean the resulting rows here.