[Community Submission] Add skill: fill-missing-product-attributes

[raqibdev]

Description

Extracts color, size, size_system, size_type, material, gender, and age_group values already present in a Shopify apparel CSV and moves them into the correct attribute columns. Produces a corrected CSV, a Markdown change log, and a needs-review file across a 3-turn merchant conversation. English catalogs only.

What's in this PR:

• Core skill (SKILL.md) with full 3-turn conversation flow, detection logic, edge cases, and error handling
• Detection script (detect_missing_attributes.py) for deterministic extraction via option values, tag prefixes, title vocabulary, body HTML, and sibling variant propagation
• Application script (apply_fills.py) merging deterministic and LLM-approved fills, writing corrected CSV, Markdown change log grouped by product, and needs-review file with priority and action columns
• Asset files: color vocab, material synonyms, gender and age patterns, apparel signal terms, size type patterns
• Reference docs: extraction source priority table, Shopify column detection logic, full 3-turn example output, Google Merchant spec reference
• Fixtures and regression test (test_fixtures.py) comparing script output against expected CSV
• Edge-case tests (test_edge_cases.py) covering XLSX rejection, empty CSV, dry-run, conflict rows, and Latin-1 encoding

Key decisions made during development:

• Detection script auto-creates a temp directory and emits work_dir in stdout JSON so downstream stages locate files without manual path management; temp dir deleted on completion
• proposed_fills.json dropped: LLM inferences held in conversation memory and written directly to approved_fills.json after merchant approval, removing a silent intermediate file between Turn 2 and Turn 3
• change_log.csv replaced with change_log.md grouped by product handle, with confidence percentages, source labels, and REVIEW/OK status per fill
• size_system and size_type added as target fields with full GS column pattern matching
• needs_review.csv expanded with priority (HIGH/MEDIUM/LOW) and action columns mapping internal reason codes to merchant-facing plain English
• Excel input rejected early with a clear re-export instruction rather than a cryptic parse error
• Skill reframed from "fill missing Google Shopping attributes" to "extract and consolidate values already in the CSV", matching what the script actually does

Contributor Attestation

• I agree to the Terms of Service and Privacy Policy, and I understand that if my submission is accepted it will be published under the Apache 2.0 license. I confirm I have the rights to submit this content and it does not include confidential information or personal data.