vendor-inventory-report
Skill8 starsChanged 45 days agoMITView on GitHub
Build an Amazon Vendor Central inventory report as a live dashboard - sellable and unsellable units, open purchase-order units, net received, aged 90+ days, sellable inventory cost and unfilled customer orders, with a Sellable vs Open PO chart and a sortable, filterable ASIN table - from DataDoe's Vendor Central data. In Claude Cowork it is a live artifact with a Refresh button; elsewhere a self-contained HTML file; both can be saved as PDF for stakeholders. Use for "vendor inventory report", "vendor central inventory", "1P inventory dashboard", "open PO units", "aged inventory vendor", "sell-through by ASIN", or "vendor stock report". Then offers an updated version with product names, weeks of cover per ASIN and stockout / overstock flags.
In this repository26 other files it gives its agents.
- amazon-asin-search-auditorSkill
- amazon-listing-optimizerSkill
- buy-box-loss-root-causeSkill
- create-amazon-reconciliation-dashboardSkill
- create-orders-managerSkill
- daily-account-health-checkSkill
- daily-performance-health-checkSkill
- fba-inventory-ledger-reportSkill
- fba-stock-restock-plannerSkill
- keyword-rank-sqp-trackerSkill
- net-profit-pl-analyzerSkill
- ppc-asin-acos-alertSkill
- ppc-bid-optimizer-applySkill
- ppc-campaign-acos-alertSkill
- ppc-negative-keyword-applierSkill
- ppc-profit-bid-applySkill
- ppc-wasted-spend-watchdogSkill
- promotions-managerSkill
- restock-priority-alertSkill
- return-refund-analyzerSkill
- sales-movers-scannerSkill
- suppressed-inactive-listings-checkSkill
- vendor-performance-dashboardSkill
- vendor-pnl-dashboardSkill
- weekly-business-reviewSkill
- weekly-sales-briefingSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 66
Treat null numbers as 0 unless a step says otherwise.
- Distinctive: line 132
Never put a DataDoe key, MCP URL key or Anthropic API key into the page.
- Distinctive: line 206
No API key or MCP key anywhere in the HTML?
- Distinctive: line 226
Read-only. Vendor Central only - for Seller Central FBA stock, use FBA Stock & Restock Planner.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Vendor Inventory Report
A simple but powerful ASIN-level view of your Amazon vendor (1P) inventory: what Amazon holds sellable and unsellable, what is still on open purchase orders, what was received, what is ageing past 90 days, what the sellable stock is worth, and how fast it sells through. One dashboard, refreshed from your Vendor Central data in DataDoe - cleaned and aggregated, so the assistant builds it fast and without burning tokens. Live from DataDoe.
When to use this
- Weekly (or before a vendor call) inventory check on Vendor Central.
- Spotting ASINs with stock but no open POs, or open POs on slow sellers.
- Watching aged 90+ inventory before Amazon pushes back.
- Trigger phrases: "vendor inventory report", "vendor central inventory", "1P inventory dashboard", "open PO units", "aged inventory vendor", "sell-through by ASIN".
The framework. Snapshots vs flows
Vendor inventory numbers come in two kinds - mixing them up is the classic mistake:
- Snapshots (state on one day): sellable units, unsellable units, open PO units, aged 90+ units, sellable inventory cost, unfilled customer-ordered units. Read them from the latest available date only - never sum them across days (30 daily snapshots summed = ~30x the real stock).
- Flows (movement over the window): net received units, shipped units, customer returns. Sum these over the 30-day window.
- Sell-through for the window = (shipped units - customer returns) / (sellable on hand at the start of the window + net received in the window), shown as %. Using the start-of- window stock matched Amazon's own daily sell-through rate on 116 of 116 checks; end-of-window stock pushes fast sellers above 100%.
Configuration
- MCP base:
https://mcp.datadoe.com/mcp/v1 - Export rules (every export in this skill): each
exports_createcall needslimitandoutputType. Aggregation aliases must differ from column names (sumofad_spend-> aliastotal_ad_spend; reusing the column name fails with ALIAS_COLLISION). Text columns can't be aggregated (maxon a name fails with INVALID_AGGREGATE) - put them ingroupByand merge rows in code. Load every page ofexports_source_get(a table's columns are spread over several pages). Ifexports_createalready returnsCOMPLETED, skip polling; otherwise pollexports_getevery ~5 s. For results you process in code, useexports_raw_url_getand save the file (the URL expires after ~15 minutes) instead of pulling it into the conversation. Treat null numbers as 0 unless a step says otherwise. Marketplace codes are ISO (the UK isGB). - Data source:
Sales, Traffic & Inventory by ASIN & Date(amazon_vendor_sales_traffic_and_inventory_by_child_and_date), one row perchild_asinanddate. Use the manufacturing view columns (manufacturing_retail_*); thesourcing_retail_*columns are an overlapping view of the same facts - pick one view, never add the two.- Snapshot columns:
manufacturing_retail_sellable_on_hand_inventory_units,manufacturing_retail_unsellable_on_hand_inventory_units,manufacturing_retail_open_purchase_order_units,manufacturing_retail_aged_90_plus_days_sellable_inventory_units,manufacturing_retail_sellable_on_hand_inventory_cost(+manufacturing_retail_sellable_on_hand_inventory_cost_currency),manufacturing_retail_unfilled_customer_ordered_units. - Flow columns:
manufacturing_retail_net_received_inventory_units,manufacturing_retail_shipped_units,manufacturing_retail_customer_returns. - The ASIN column is
child_asin(older builds usedasin, which no longer exists).
- Snapshot columns:
requiresDatePeriodis true: always sendfrom/to(YYYY-MM-DD).- Data lag: vendor data arrives ~4 days behind. Anchor the window on the newest
datethat has a populated inventory snapshot (not just rows), not on today, and show that date as "Data as of". - Window: last 30 days ending at that newest date (adjustable).
- Currency: from the
*_currencycolumn; never hardcode a symbol. - Row caps: 1,000 rows per JSON export, 5,000 per CSV. Pull every ASIN (CSV if large), compute totals over all of them, and embed all of them in the page.
- Missing is not zero: the inventory columns are nullable. For snapshot columns, an ASIN
with no value on
Dis unknown - show "-", leave it out of the stock totals' denominators as needed, and never flag it out of stock. Flow columns (received, shipped, returns) treat null as 0. Drop ASINs with no data at all in either export; treat a negative value (e.g. open PO -1) as 0.
Step-by-step workflow (MCP-native)
sellers_and_vendors_list-> pick the vendor account (accountType = VENDOR); ask if there are several.exports_sources_get(query "vendor sales") -> resolveamazon_vendor_sales_traffic_and_inventory_by_child_and_date;exports_source_getto confirm the columns above (page through all column pages).- Newest snapshot date:
exports_create,from= 14 days ago,to= today,groupBy ["date"],countofmanufacturing_retail_sellable_on_hand_inventory_units(aliassnapshot_rows- counts only rows where the snapshot is populated) andcountofchild_asin(all_rows), JSON.D= the newest date whosesnapshot_rowsis close to its usual level. This table joins sales/traffic and inventory, so a date can have ASIN rows with no inventory snapshot at all - picking it by row count would read every stock figure as 0. - Snapshot export:
from=to=D,groupBy ["child_asin", "manufacturing_retail_sellable_on_hand_inventory_cost_currency"], sum each snapshot column (aliasessellable,unsellable,open_po,aged_90,sellable_cost,unfilled), CSVlimit 5000. - Start-of-window stock: the same sellable sum for the day before the window starts
(
D- 30; if that date has no rows, the nearest earlier date),groupBy ["child_asin"], aliassellable_start- used only for sell-through. - Flow export:
from=D- 29 days,to=D,groupBy ["child_asin"], summanufacturing_retail_net_received_inventory_units(net_received),manufacturing_retail_shipped_units(shipped),manufacturing_retail_customer_returns(returns), CSVlimit 5000. - Download each export (
exports_raw_url_getand save the file when you build the page in code). Merge onchild_asin; compute sell-through per ASIN and for the total; KPIs over all ASINs; default sort bysellabledesc. - Build the dashboard (Output format) for the client you are in:
- Claude Cowork: a live artifact. Embed the current data so it renders instantly, and
wire the Refresh button to re-run steps 3-7 through the Cowork runtime
(
window.cowork.callMcpTool('<the DataDoe exports_create / exports_get / exports_raw_download tool names exactly as in your tool list>', args)), declaring those three tools in the artifact's MCP tool list. Parse the MCP envelope (content[0].text-> JSON; the id field isexportId). - Any other client (Claude Code, Claude.ai, ChatGPT, Cursor): one self-contained HTML file with the data embedded; refresh = run the skill again.
- Never put a DataDoe key, MCP URL key or Anthropic API key into the page. Outside Cowork the Refresh button shows "Live refresh works inside Claude Cowork - re-run the skill to update" instead of asking for a key.
- Claude Cowork: a live artifact. Embed the current data so it renders instantly, and
wire the Refresh button to re-run steps 3-7 through the Cowork runtime
(
- Save, report the path (or show the artifact), and offer the PDF.
- Offer the updated version (see below): "There's an updated version of this report - it adds product names, weeks of cover per ASIN and flags for stockout risk and overstock. Want it?" Build it only on a yes.
Output format
Dark theme (background #0A0F1E, cards #111827, accent DataDoe green #00C896, Inter /
system font), Chart.js 4 from a CDN:
- Header: "Vendor Inventory" · vendor name + marketplace · "Last 30 days ({from} - {to})" · "Data as of {D}" · Refresh button · Download PDF button.
- KPI cards (7): Total Sellable Units · Total Unsellable Units · Open PO Units · Net Received (30d) · Aged 90+ Days · Sellable Cost ({currency}) · Unfilled Customer Orders.
- Chart: "Sellable vs Open PO by ASIN" - grouped bars, top 15-20 ASINs by sellable.
- ASIN-level detail table: ASIN · Sellable · Unsellable · Open PO · Net Received · Aged 90+ · Unfilled Orders · Sell-through % · Cost. All ASINs are embedded; the table shows the first 50 rows of the current sort / filter (default: top 50 by sellable) with "show all". Sorting and the ASIN filter box work on every ASIN, so an out-of-stock ASIN is never hidden below the top 50. Highlight aged 90+ > 0 (amber) and sellable = 0 with open PO = 0 (red - nothing coming). A total row at the bottom carries the all-ASIN totals and the total sell-through.
- Footer: "Data via DataDoe".
PDF: the Download PDF button calls window.print() with print CSS (A4 landscape, light
background, full table) so the user can send the snapshot to a client or stakeholder. Offer an
optional weekly routine (scheduled task) that regenerates the report and emails the PDF to the
recipients the user names.
Loading and errors (live artifact): show a spinner during exports; on failure keep the last data on screen with a clear banner ("Refresh failed - showing data as of {D}") instead of blanking the page.
Updated version (offer after the standard report)
Same header, KPIs, chart and table - the updated version adds what to do about the stock. Build it only when the user says yes (the Refresh button then re-runs the extra exports too).
- Product Name column next to ASIN, from
Product Catalog by ASIN(amazon_products_by_child_asin, vendor catalog:child_asin,product_name; no date range needed). Filterchild_asin in <the report's ASINs>(comma-separated; split long lists) - an unfiltered pull hits the 5,000-row cap because titles repeat per language - and keep one title per ASIN (the marketplace's language if you can tell, else the first), cut to ~60 characters. ASINs still missing a title keep the ASIN only. - Weeks of Cover = sellable units / weekly demand, where weekly demand = 30-day
manufacturing_retail_ordered_units/ 30 x 7 (add it to the flow export). "-" when there was no demand. - Flag column (and a KPI card "ASINs at risk" = 🔴 + 🟠):
- 🔴 Out of stock - sellable reported as 0 (not unknown) with customer demand in the window.
- ❔ No stock data - the snapshot is missing for this ASIN on
D; never treated as 0. - 🟠 Stockout risk - under 4 weeks of cover and 0 open PO units.
- 🟡 Overstock - over 26 weeks of cover, or aged 90+ above 20% of sellable with more than 8 weeks of cover; also stock with no demand at all ("no demand").
- ✅ OK otherwise. Thresholds are adjustable.
- A "Flag" filter above the table so the user can show only the ASINs that need action.
Worked example (illustrative)
German vendor account, data as of 13 May (today 17 May). 412 ASINs; KPIs: 38,950 sellable, 612 unsellable, 9,840 open PO units, 21,300 net received, 2,150 aged 90+, €214,600 sellable cost, 180 unfilled. ASIN A tops the table with 4,120 sellable, 0 open PO and 61% sell-through - healthy now, but nothing on order. ASIN B: 900 sellable, 610 of them aged 90+, 4% sell-through
- an overstock conversation for the next vendor call.
Quality self-check
- Did I take snapshot columns from one date (the newest with data) and sum only the flow columns over 30 days? Sell-through on start-of-window stock?
- Did I anchor the window on the newest data date and show "Data as of"?
- Are the KPIs totals over all ASINs, not just the 50 in the table?
- Did I use
child_asinand only the manufacturing view? - Is the currency taken from the data?
- No API key or MCP key anywhere in the HTML?
- Did I offer the updated version at the end - and change nothing unless the user said yes?
Common mistakes
- Summing sellable / open PO / aged units over 30 daily rows (inflates stock ~30x - 30.05x in testing).
- Grouping by
asin- the column ischild_asin. - Picking the snapshot date by row count, or treating a missing snapshot as 0 - stock reads 0 and ASINs get flagged out of stock.
- Adding
manufacturing_retail_*andsourcing_retail_*values together (double count). - Ending the window at today - the last ~4 days are empty for vendors.
- JSON
limitabove 1,000 (the export fails) - use CSV for large catalogs. - Computing KPIs from the top-50 table only, or embedding only the top 50 (out-of-stock ASINs are never in a top 50 by sellable).
- Sell-through on end-of-window stock (fast sellers go above 100%) - use start-of-window stock.
- Embedding credentials to make "Refresh" work in a normal browser.
Notes
- Read-only. Vendor Central only - for Seller Central FBA stock, use FBA Stock & Restock Planner.
- Sourcing-view metrics (PO fill rates, contra-COGS) live in the
sourcing_retail_*columns of the same table if you need them later. - A DataDoe skill, built on
amazon_vendor_sales_traffic_and_inventory_by_child_and_date.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

