For Amazon FBA & Shopify sellers · v0.5

Find the cash
hiding in your stock.

stockcash is an MCP server for sellers doing $10k–$250k/month with 100–1,500 SKUs. It reads the inventory and order exports you already have, then tells your agent how much cash is stuck, what to mark down, what to reorder, and what changed since last week. Hosted URL or local.

Free tier · no card · hosted URL or local stdio
analyze_inventory90d · USD
# two csv files, deterministic math cash tied $75,124 dead stock $2,100 2 SKUs slow stock $17,479 3 SKUs overstock $37,844 14 SKUs reorder capital $26,551 net cash plan +$95,721 ## Do this week 1. markdown ladder on 19 SKUs 2. 2 urgent reorders 3. bundle FT-208 + HM-102
Works with any MCP client
ClaudeClaude CodeCursorCodexOpenCodeWindsurfClineZed
Who it is for

Built for sellers with cash on the shelf.

Not a generic analytics tool. It is shaped around how small e-commerce teams actually buy stock: 30–60 days ahead, then wait and hope the cash comes back.

AMAZON FBA

Solo sellers, 100–1,500 SKUs

You export an inventory ledger and order report. stockcash finds the SKUs that stopped selling, shows how deep you can discount before cost, and what restock capital is really needed.

SHOPIFY / DTC

Seasonal buys and slow movers

Product and order exports are enough. See which variants are overstocked, what a bundle can absorb, and the net cash plan before you commit to the next buy.

AGENCIES / VAs

Running 5–30 stores

Save a profile per store, run a weekly digest, and open one portfolio table with the single best move for every client. No client login required.

Not for dropshippers with no owned inventory, or teams that need ERP / 3PL integration. If you don’t export CSVs, this will not fit.

Why

Dashboards report. This decides.

Sellers do not need another chart. They have stock and no cash. The useful answer is which SKUs to discount, how deep before selling below cost, what to reorder, and which dead units a best seller can carry.

01

Two files, no integrations

Amazon, Shopify, Etsy, Walmart or a hand-kept sheet. If it exports CSV, it works. No OAuth, no keys, no platform ToS surface.

02

Formula on the page

Velocity, cover, break-even discount and reorder quantity are arithmetic on your own rows. Any number can be checked by hand.

03

Memory a chat lacks

Profiles, snapshots and a weekly diff. It separates “just stopped selling” from “dead for a year”, and shows what changed.

Tools

13 tools. One server.

ToolWhat it answersCredits
analyze_inventoryWhere is my cash stuck, and what do I do this week?2
cash_summaryHow much cash is tied, in one screen?1
dead_stock_reportWhat is not selling, and how deep can I discount?2
reorder_planWhat do I buy, when, and with how much cash?2
bundle_ideasHow do I clear dead stock without giving it away?2
trend_reportWhat just stopped selling? What is recovering?1
snapshot · weekly_digestWhat changed since last week?1 · 3
portfolio_digestEvery store in one table.5
validate_exports · demo_report · licenseFirst run, no files, plan status.0
Setup

Connect in 30 seconds.

Hosted: sign up, copy your API key, paste one URL. Local: clone and run stdio. Same 13 tools either way.

STEP 01

Create an account

Free plan, no card. Your dashboard issues an API key and shows the hosted endpoint.

STEP 02

Export two CSVs

Inventory (SKU, on hand, cost, price) and orders (date, SKU, units). Column aliases are detected.

STEP 03

Ask for the decision

“Where is my cash stuck and what should I do this week?” The agent calls the tools.

hosted · one URLcli
# copy the key from your dashboard claude mcp add --transport http stockcash \ https://mcp.yourdomain.com/mcp \ --header "Authorization: Bearer sc_live_..."
local · stdiojson
{ "mcpServers": { "stockcash": { "command": "node", "args": ["/path/to/stockcash-mcp/dist/index.js"] } } }
No data yet?

Try the demo store first

Ask for demo_report. It builds a deterministic 32-SKU store and prints the full report, so you can judge the output before exporting anything. No account needed.

One plan

Free to try.
$19 when it remembers.

The paid boundary is state. Static reports are free on up to 50 SKUs. Pro adds snapshots, the weekly digest, the portfolio view, unlimited SKUs and 1,000 credits a month.

free 20 credits · 50 SKUs · 1 store pro $19/mo · 1,000 credits · unlimited gap snapshots · digest · portfolio hosted mcp.yourdomain.com · key metered local signed key · verified offline
FAQ

Questions sellers ask.

Who is this for?

Small Amazon FBA and Shopify / DTC sellers doing roughly $10k–$250k a month with 100–1,500 SKUs, plus agencies and VAs running several stores. It is built for people who buy inventory 30–60 days ahead and feel the cash gap. It is not for dropshippers or teams needing ERP/3PL integration.

Do you get access to my store or data?

Local mode: no. The server reads CSVs you point it at and makes zero network calls; profiles and snapshots stay in ~/.stockcash-mcp/. Hosted mode: your CSVs are sent over TLS to the hosted endpoint so it can answer, and the gateway meters credits. Choose local if the data must never leave your machine.

Which platforms work?

Any export with SKU, quantity, cost and price: Amazon, Shopify, Etsy, WooCommerce, Walmart, TikTok Shop or a spreadsheet. Aliases like seller_sku, MSKU, ASIN, AFN Fulfillable Quantity and cogs are detected.

What is a credit?

One successful tool call. Quick calls cost 1, reports 2, a weekly digest 3, a portfolio digest 5. demo_report, validate_exports, store_profile and license are free. Failed calls cost nothing.

Do I need an account?

Only for the hosted URL. Sign up, copy the API key, paste one endpoint. The local build needs no account: a signed license key is verified offline and works without a login.

Hosted or local — which should I pick?

Hosted is a one-line install and metered server-side, so it is the easiest path and works from any machine. Local is a single binary on your machine, makes zero network calls, and is the right choice if exports must never leave your infrastructure. Both run the same 13 tools.

Does it model price elasticity?

No, and the report says so. It shows the markdown ladder and the lift you would need. You step one tier at a time and re-run.

Two CSVs. One clear answer.

Start on the free plan. Upgrade only when you want it to remember.