Your AI-powered sourcing assistant for Amazon FBA resellers.
ArbitrageOps runs inside Claude and handles your full arbitrage workflow — finding competitor storefronts, scoring Keepa data for buy signals, calculating true cost with discount stacking, and tracking your pipeline from lead to sold. No spreadsheet juggling, no copy-pasting between tabs.
Designed for resellers, not developers. If you can download a folder and open an app, you're ready.
| Requirement | Notes |
|---|---|
| Claude Desktop | Free download — Mac or Windows |
| Claude Pro subscription | Required for scheduled tasks and full workflow use |
| Google account or Microsoft 365 | For your pipeline spreadsheet (Google Sheets recommended) |
| Chrome browser | For the optional Chrome extension |
Mac or Windows (no technical skills needed):
- Click the green Code button at the top of this page
- Click Download ZIP
- Unzip the folder — put it somewhere easy to find (Desktop or Documents)
If you use Git:
git clone https://github.com/Po1o85718/arbitrage-ops.git
Download and install Claude Desktop from claude.ai/download if you haven't already. Sign in with your Claude Pro account.
- Open Claude Desktop
- Click the Cowork tab
- Click Projects, then New Project (or open an existing one)
- Click Choose Folder and navigate to the
arbitrage-opsfolder you unzipped — then click Open (Mac) or Select Folder (Windows) - Click Save
Claude Desktop will load ArbitrageOps automatically every time you open it from now on.
Stuck? See Troubleshooting → "Claude doesn't recognize /arbitrage-ops" for more options.
Type this in Claude and press Enter:
/arbitrage-ops setup
Claude will walk you through everything — your ROI targets, where to store your pipeline data (Google Sheets or Excel), the Chrome extension, and your daily morning briefing. Setup takes about 5–10 minutes.
Once setup is complete, these are the commands you'll use every day:
| Command | What it does |
|---|---|
/arbitrage-ops |
View your pipeline dashboard — leads, purchases, live inventory, profit |
/arbitrage-ops hunt |
Find ASINs by stalking a competitor's storefront (Amazon or Mercari) |
/arbitrage-ops analyze |
Score a Keepa CSV or liquidation manifest for buy signals |
/arbitrage-ops source |
Calculate your true cost after stacking all discounts |
/arbitrage-ops log |
Record a lead or purchase into your pipeline |
/arbitrage-ops setup |
Return here anytime to change your settings |
You can also just paste a storefront URL or drop a CSV — Claude auto-detects and runs the right workflow.
ArbitrageOps looks for a specific pattern called a Supply Squeeze — when Amazon sellers are dropping out of a listing while the price is rising. That's the signal to buy.
A product scores as a buy when:
- Seller count has dropped 10%+ in the last 30 days
- Buy Box price is up 5%+ vs the 90-day average
- Best Sellers Rank is under 100,000
- Amazon itself is out of stock more than 70% of the time
Buy Strength Score (1–10):
| Score | Signal | Action |
|---|---|---|
| 8–10 | Strong squeeze, 35%+ ROI | Buy |
| 5–7 | Moderate signal, 25–35% ROI | Investigate further |
| 1–4 | Weak signal | Pass |
ROI Formula:
Net Profit = (Buy Box Price × 0.85) − True Buy Cost − FBA Fees
ROI % = Net Profit ÷ True Buy Cost × 100
Minimum target: 30% ROI. Never commit more than $500 to an untested ASIN.
ArbitrageOps calculates your true cost after stacking every available discount:
| Layer | Tool | Typical Savings |
|---|---|---|
| 1. Cashback portal | Rakuten or Capital One Shopping | 3–10% |
| 2. Discounted gift card | Raise.com | 2–5% |
| 3. Coupon at checkout | Honey | varies |
| 4. Credit card rewards | 2% cashback business card | 2% |
| Combined | ~10–17% effective savings |
Set up HARPA AI to monitor Raise.com for gift card price drops on your go-to retailers — it runs in the background and emails you when a card hits your target discount.
ArbitrageOps supports three types of sourcing:
Retail / Online Arbitrage — buy from retailers (Target, Walmart, Home Depot, etc.) using the discount stack above, sell on Amazon FBA.
Mercari Storefront Stalking — find resellers on Mercari who are pricing brand-new items below Amazon value. Buy from them, relist on FBA.
Liquidation Lots — buy pallets or cases of customer returns:
- Direct Liquidation — electronics lots (higher margin, higher risk)
- BULQ — Amazon return cases (household, toys, baby — great starter lots)
These aren't required to start, but they make your workflow significantly faster:
| Tool | What it does | Cost |
|---|---|---|
| Keepa | Amazon price & sales history tracker | ~$19/mo |
| Seller Assistant | Batch ASIN research + Storefront Widget | ~$15/mo |
| SellerAmp SAS | Real-time profit calculator & restriction checker | ~$19/mo |
| Tactical Arbitrage | Automated product scanning across 1,000+ stores | ~$89/mo |
ArbitrageOps tracks every lead through five stages:
PENDING → SOURCED → INBOUND → LIVE → SOLD_OUT
If inventory stalls, Claude flags it as STRANDED and walks you through recovery options (Mercari, eBay, local resale, or an Amazon price drop).
Your pipeline data lives in your Google Sheet or Excel workbook — created automatically during setup. No files to manage.
- Morning — Claude sends your briefing (scheduled task you set up during setup): pipeline snapshot, flags, and top 3 leads to act on today
- Sourcing — paste a storefront URL →
/arbitrage-ops hunt→ get an ASIN list → run through Keepa →/arbitrage-ops analyze - Buying —
/arbitrage-ops source→ discount stack → true cost → buy decision - Logging —
/arbitrage-ops log→ purchase recorded to your sheet automatically - Reporting —
/arbitrage-opsanytime for a full pipeline dashboard
ArbitrageOps needs access to your Google Sheet (or OneDrive) and optionally Keepa.
These credentials are stored inside the project folder in a .env file — never committed to GitHub.
The setup wizard creates this file for you, but here's what it looks like:
ARBITRAGE_DATA_BACKEND=sheets
GOOGLE_SHEETS_ID=your-sheet-id-here
KEEPA_API_KEY=your-keepa-api-key-here # optional
Copy .env.example to .env to get started. The .env file is gitignored — your keys stay local.
Advanced users: You can also put credentials in
.claude/settings.local.json(see.claude/settings.local.example.jsonfor the format). Both work the same way.
arbitrage-ops/
├── CLAUDE.md # Loaded by Claude on every session
├── .env.example # Credential template (copy to .env)
├── .claude/
│ ├── settings.json # Model config and permissions
│ ├── settings.local.example.json # Credential template for settings approach
│ └── commands/ # Slash command definitions
├── config/
│ ├── profile.example.yml # Template — copied to profile.yml during setup
│ └── profile.yml # Your live settings (gitignored)
├── modes/
│ ├── _shared.md # Scoring rules, ROI formula, data routing
│ ├── analyze.md # Keepa CSV + liquidation manifest scoring
│ ├── hunt.md # Storefront stalking (Amazon, Mercari, liquidation)
│ ├── source.md # Discount stack + liquidation buying process
│ ├── log.md # Pipeline tracker + stranded recovery
│ └── report.md # Dashboard + command dispatcher
├── templates/ # Column definitions for Sheets/Excel setup
├── examples/
│ └── keepa-sample.csv # Sample Keepa export to test /arbitrage-ops analyze
└── docs/ # Design, how-it-works, troubleshooting guides
config/profile.ymland.envare gitignored — they stay on your computer only- Pipeline data lives in your own Google Sheet or OneDrive — ArbitrageOps never touches it
- No credentials are stored in the repo — only in your local
.envorsettings.local.json
Built for Claude Pro on Mac and Windows. Requires Claude Desktop.