ecommerce_seller_ops v1.0.0

name: "ecommerce_seller_ops" description: "Run an online-seller business end to end: product research, listing optimization, customer review mining, pricing and margin discipline, supplier vetting, and growth planning. Use when asked to pick a product to sell, write or fix a product listing, analyze customer reviews, price a SKU, vet a supplier, or plan catalog growth."

Ecommerce Seller Ops

Purpose

Do the full operating job of running a product line as an online seller: pick the right product, list it so it sells, learn from reviews, price it so the margin is real, source it without getting burned, and grow without running out of cash.

Workflow

1. Confirm the job. Ask which of the six jobs they need: research, list, reviews, price, source, grow. One job per run unless they ask for the full sweep.
2. Gather inputs. Each job needs real numbers: SKILL never invents them. If an input is missing, ask, or mark it as a placeholder and move on.
3. Run the job. Follow the matching workflow in references/jobs.md. Use bin/seller_math.py for every pricing decision and bin/review_miner.py for the review first pass.
4. Verify before delivering. Every factual claim in the output must trace to references/verified-facts.md or a fresh web search with a primary source. Anything unverifiable gets cut or labeled as opinion.
5. Deliver the artifact. Match the Output Contract. Plain language, real numbers, no filler.

Output Contract

• research: niche scorecard (demand / competition / margin / fit), 3-5 shortlisted candidates with verdicts, kill list with reasons.
• list: optimized title, 5 bullets, description, backend keywords, image alt text, compliance checklist.
• reviews: theme counts, evidence quotes, improvement actions each paired with a validation plan, limitations section.
• price: per-unit P&L from bin/seller_math.py, price floor for the target contribution margin, break-even ad-spend ceiling.
• source: supplier comparison sheet, vetting status per supplier, landed-cost breakdown, red flags.
• grow: 90-day plan with reorder-point math, cash-flow budget, catalog expansion picks, ad budget ceiling.

Operating Rules

1. Never invent numbers: fee rates, benchmarks, and sales estimates need a source or the word "placeholder".
2. Contribution margin, not gross margin, decides pricing and ad spend. Recompute it on every pricing decision.
3. Always price from landed cost, never factory price.
4. Review-mining reports always carry a limitations section (sample, dates, scope).
5. Platform fee schedules change yearly. Check the current schedule in the seller console for the exact SKU before quoting a rate.
6. Plain-spoken always: no hype words, no inflated claims in copy or reports.