version: "1.0.0" name: sales-analysis description: Turn the RAW weekly sales rows and catalog facts from the Sales & Performance MCP into merchandising analytics by running the bundled sales_analysis.py script. Use first in an end-of-quarter markdown review, when a manager wants top sellers and slow-moving / dead stock over a period. The MCP returns raw data only, so the script derives velocity, weeks-of-cover, and margin, then prints the top sellers and slow movers as structured JSON. It renders NO charts (the merch-report-html skill owns all visuals) and does NOT decide discounts (hand the slow movers to markdown-optimizer).

Sales Analysis (data, not charts)

This skill converts raw sales data into the analytics a markdown review needs by running the bundled `sales_analysis.py` script. The Sales & Performance MCP deliberately exposes only raw primitives (query_sales, get_catalog) - there is no get_slow_movers or get_velocity - so the script derives the metrics itself.

It renders no charts on purpose: the interactive merch-report-html skill owns all the visuals at the end, which keeps the review fast. This skill just computes the numbers and prints the structured hand-offs.

When to use

First in an end-of-quarter markdown review, whenever a manager wants to see top sellers and slow / dead stock over a window.

Inputs (fetch from the Sales & Performance Agent, then pass as flags)

1. Raw weekly rows - query_sales(start_week, end_week) for the window (default the

full 8 weeks 2026-04-12 ... 2026-05-31). Each row is {sku, name, category, week, units, revenue}.

1. Catalog facts - get_catalog() -> {sku, name, price, unit_cost, stock_on_hand}.

Run the bundled script

python3 sales_analysis.py \
    --woc-threshold 8 \
    --rows '[{"sku":"SKU-VR-GOGGLES","name":"OmegaVision VR Headset","category":"accessory","week":"2026-04-12","units":1,"revenue":199.99}]' \
    --catalog '[{"sku":"SKU-VR-GOGGLES","name":"OmegaVision VR Headset","price":199.99,"unit_cost":70.00,"stock_on_hand":48}]'

It prints a compact summary, then two structured lines:

• TOP_SELLERS_JSON=[{name, units, velocity, revenue}]
• SLOW_MOVERS_JSON=[{sku, name, price, cost, weeks_of_cover, margin, units}]

Metrics it derives (for your report, not for you to recompute)

Over a window of W weeks, per SKU: velocity = units_total / W, weeks_of_cover = stock_on_hand / velocity, margin% = (price - cost) / price * 100. Slow movers = weeks-of-cover above the threshold (default 8), sorted descending. Top sellers = highest units.

Next steps

• Pass SLOW_MOVERS_JSON (it already carries sku, name, price, cost) to the

markdown-optimizer skill, together with the guardrails from the Store Policy Agent.

• Keep TOP_SELLERS_JSON and SLOW_MOVERS_JSON so the merch-report-html skill can

render them into the final interactive report.

• Do not recommend discounts here - that needs the Store Policy Agent's guardrails.

Notes

• The bundled script is sales_analysis.py; it uses only the Python standard library

(no matplotlib, no network).

• Never invent numbers - every value derives from the MCP results you fetched. All data

is mock; for demo purposes only.