Product Data Cleanup

/as:product-data-cleanup — Product Data Normalizer

Harness note: use /as:<skill> on Claude Code and $<skill> on Codex. Resolve <skill-root> as the directory containing this loaded SKILL.md and <plugin-root> as the plugin root that contains skills/, and use equivalent native tools when host tool names differ.

Takes a messy FF&E schedule and normalizes everything: casing, dimensions, units, language, materials vocabulary, currency formatting, and duplicates. Outputs a clean, consistent, spec-ready schedule.

Persistent cleanup operates on the nearest project's product-library.csv. Pasted tables may be previewed in Markdown but are not another persistent format.

Input

The user provides a schedule in one of these ways:

1. Project library — the nearest product-library.csv under an ancestor containing PROJECT.md.
2. CSV file path — import only after its exact 33-column header is validated.
3. Pasted table — preview a proposed canonical mapping before any persistence.

If the input format is unclear, ask.

Cleanup Rules

1. Casing

Known brand abbreviations to preserve: HAY, USM, B&B, DWR, CB2, HBF, OFS, SitOnIt, 3form, ICF

2. Category Normalization

Map free-text categories to the canonical vocabulary and alias table defined in ../../schema/product-schema.md. Read that file for the full mapping of variations (English, Spanish, legacy terms) to canonical category names.

If a category is ambiguous, keep the closest match and add a [?] flag for the user to review.

3. Dimensions

Splitting combined dimensions:

Dimension rules:

• Always store as separate W, D, H columns with a Unit column
• If dimensions are already split, validate they're numeric (strip any unit text from the number)
• Interpret " as inches, ' as feet (convert to inches: 2'6" → 30)
• Accept ×, x, X, by, por as separators
• Convention: W × D × H (width × depth × height). If only 2 values, ask which is missing.
• Round to 2 decimal places max
• If unit is missing but values suggest inches (all < 100 for furniture), assume in. If values suggest cm (> 100 or explicit), use cm. If truly ambiguous, flag with [?].

Do NOT convert units. Keep the original unit. Designers need the manufacturer's spec for ordering.

4. Language Normalization

Detect the language of each field value and normalize to English unless the user specifies otherwise.

Rule: Translate category, material, and color/finish fields. Leave Product Name and Brand as-is (proper nouns).

If the user says "keep in Spanish" or specifies a target language, respect that.

5. Materials & Finishes Vocabulary

Standardize common material terms:

6. Price & Currency

• Strip currency symbols ($, €, £, ¥) — store symbol as currency code in separate column
• Remove thousands separators (both . and , — detect locale: 1.234,56 is EU format, 1,234.56 is US)
• Store as plain decimal number: 5695.00
• If price says "Contact", "Quote", "Trade", "A consultar", "Consultar" → set to empty
• Currency detection: $ alone defaults to USD unless context suggests otherwise (UY site → UYU, EU site → EUR)
• If a schedule mixes currencies, keep each row's original currency. Add a note at the top.

7. Duplicate Detection

• Flag rows with identical Product Name + Brand as potential duplicates
• Flag rows with identical URL as definite duplicates
• Don't auto-delete — present duplicates to the user and ask what to keep

8. Whitespace & Formatting

• Trim leading/trailing whitespace from all fields
• Collapse multiple spaces to single space
• Remove line breaks within field values
• Normalize list separators: wood / metal / glass → Wood, Metal, Glass (comma-separated)
• Remove trailing commas or semicolons

Workflow

Step 1: Load the schedule

Read the input. Report: "Loaded N rows with M columns." Map input columns to the canonical schema. If column mapping is ambiguous (e.g., a column called "Size" could be combined dimensions), ask the user.

Step 2: Analyze issues

Scan all rows and produce a summary:

## Cleanup Preview

- **Casing**: X product names need Title Case
- **Categories**: Y rows have non-standard categories (mapping: "chairs" → Seating, etc.)
- **Dimensions**: Z rows have combined dimensions to split
- **Language**: W rows have Spanish-language fields to translate
- **Materials**: V rows have non-standard material terms
- **Prices**: U rows need currency formatting cleanup
- **Duplicates**: T potential duplicate rows found
- **Empty fields**: S rows missing dimensions, R rows missing price

Step 3: Confirm scope

The issue summary is the change preview. Present selectable cleanup groups directly through the single confirmation gate; do not ask the same question first in prose.

Step 4: Apply fixes

Process every row through the active cleanup rules. Track every change made.

Step 5: Present results

Show a before/after diff for a sample of changed rows (up to 5 examples). Then show the full cleaned table.

Report:

## Cleanup Complete

- Rows processed: N
- Changes made: X
- Flagged for review: Y (marked with [?])

Step 6: Save

Read ../../schema/product-schema.md and ../../schema/csv-conventions.md. For multiple changed rows, materialize the complete proposed 33-column CSV as a temporary or user-visible review file, validate that candidate, preview the whole change once, and use the single confirmation gate. After approval, invoke python3 "<plugin-root>/skills/master-schedule/scripts/csv-library.py" import product --project <project-root> --source <review.csv> exactly once for one atomic replacement; never loop update. A genuinely single-record edit may instead invoke the same plugin-root helper's update command once with one uniquely matching stable field. Never overwrite an arbitrary input or hand-edit product-library.csv.

Edge Cases

• Mixed-language schedule: Detect dominant language per column, normalize to one language
• Merged cells or irregular formatting: Flag and ask user how to handle
• Extra columns not in schema: Reject persistence and preview how they would map to canonical fields
• Empty rows: Remove silently
• Header detection: Auto-detect header row (first row with text that matches known field names). If uncertain, ask.