Product Research · from an X post, Nov 9, 2026 · Added
Operator OS Folder for Claude Code and Codex
Run Amazon brand work from one folder of rules that Claude Code and Codex both read, connected to seller tools through MCPs
He says the biggest unlocks were working in Claude Code, having the right MCPs connected, and working in a GitHub-backed folder so the AI always has context.
Per source Self-reported by the source, not independently verified.
Copy-paste prompt
Open my operator folder (operating-principles.md, brand-guidelines.md). Using the keyword and rank data available through my connected tools: 1. List keywords where my conversion rate is above the market average but my organic rank is worse than page 1. 2. For each, suggest a ranking push (bid, placement modifier, budget) that keeps the product at a 20% margin after ad spend. 3. Show the math (CPC x clicks per order = cost per acquisition) and flag anything you couldn't verify. Don't change any campaign; give me the list to approve.
The source didn't publish its exact prompt. We wrote this starter so you can try the same workflow. Replace anything in [brackets].
What they did
Evan Swanson, who says he manages brands doing $60M a year and is building three of his own, works in Claude Code and Codex instead of chat. Both open the same 'operator OS' folder holding his Amazon operating principles and brand guidelines, backed up to GitHub, so he can switch between them without losing context. Through MCPs he connects Gmail and Google Drive, Slack, Higgsfield for images and video, SoldScope (market research, rank tracking, catalog health alerts), SellerSprite (keyword conversion data), Scale Insights (PPC), Canva, BrandSearch, Shopify, Sellerise (Amazon P&L) and GitHub. His example asks include: find keywords above market conversion rate that rank poorly and suggest ranking pushes, suggest placement modifiers and bids to reach a 20% margin, estimate cost per acquisition for a new keyword set from CPC and conversion rate, and reply to a supplier email using the project folder.
How it went
He says the biggest unlocks were working in Claude Code, having the right MCPs connected, and working in a GitHub-backed folder so the AI always has context. He shares the setup and example prompts, not performance numbers.
We only report what the source says. Results are self-reported unless noted.
Worth knowing
X post (Nov 9, 2026). The tool list mixes Amazon and off-Amazon tools; the Amazon parts are PPC, keyword, P&L and catalog work. He notes Scale Insights may be replaced by a direct hookup through Amazon's official Ads MCP.