Skip to content
AEOsome
Log inSign up

MCP server and CLI for ecommerce AI visibility

Ask Claude, Cursor or your own scripts which ChatGPT and Google AI Mode shopping answers name your store, who they name instead, and what to fix first. For Shopify and WooCommerce merchants and the agencies that run their stores.

Compare plans with API accessRead the setup guide

A real session, start to finish

The command line first, then Claude reading the same data through the MCP server.

  1. aeosome whoami

    Workspace  Halden OutfittersPlan       Pro T1 (Lifetime)

  2. aeosome dashboard

    Period (days)                       30AI visibility (% of answers)        60AI visibility change (points)       0Top-3 presence (% of answers)       13.7Share of voice (%)                  21.8Shopping appearance (% of answers)  29.7Answers analysed                    300Products tracked                    10

  3. aeosome competitors

    Period (days)     30Answers analysed  300 Brand              You  Share of voice (%)  Questions won  Citation share (%)Ridgeline          no   24.7                17             0Halden Outfitters  yes  21.8                37             13.7Stridewell         no   20.6                12             0Calder Athletic    no   16.5                17             0Northmoor          no   16.5                6              0

  4. aeosome prompts --missing-brand true --json | jq -r '.items[] | select(.topRival) | [.queryText, .topRival.name] | @tsv'

    best shoes right now	Calder Athleticmy old shoes finally gave out — what should I look at?	Calder Athletictop rated shoes for everyday use	Calder Athletic

  5. aeosome fixes --priority high --json | jq -r '.items[].recommendationText'

    No llms.txt file - AI agents cannot discover site information

  6. claude -p 'Which shopping questions leave me out, and what should I fix first?'

    Six shopper questions leave Halden Outfitters out, all with 0% visibility and 100 runs between them. Four are shoe-category questions: "best shoes right now" (20 runs), "top rated shoes for everyday use" (20), and "my old shoes finally gave out" (20) name Calder Athletic as the top rival, and "shoes that lasts longer than mine did" (20) names no rival in the data. The two budget questions, "good shoes under $89.00" and "good shoes under $98.00" (10 runs each), also name no rival. I haven't read the full answers, so I can't say who fills those slots. Across the store, AI visibility is flat at 60% (180 of 300 prompts), but top-3 presence fell from 16.7% to 13.7%, share of voice from 25.5% to 21.8%, and Ridgeline now leads at 24.7% against Calder's 16.5%. Fix first: the FAQ on the product behind "good shoes under $89.00" (b1c29d33, minutes of work), because it is the only pending fix that names a question you're missing. Then add the high-priority llms.txt (impact score 80), which isn't tied to any specific question. The data covers only five days with runs, so treat these shares as early readings.
Recorded 11 October 2026 against Halden Outfitters, the fictional store we run for demos. Colour codes removed, nothing else changed.

Built for the way stores and agencies already work

  • Store owners

    Your Monday brief, in Claude

    Ask for the week in one paragraph: what slipped, which rival was named instead, and what changed.

    aeosome brief
  • Agencies

    One report per client, scripted

    Create a key in each client's workspace and pull every number as JSON into the report you already send.

    aeosome competitors --json
  • Developers

    Alerts from a scheduled job

    Send new visibility drops and rivals to Slack or email. Each failure exits with its own code, so a revoked key never passes silently.

    aeosome alerts --json
  • Content editors

    Product pages, fixed from Cursor

    While you edit a product page, ask which shopper questions leave it out and which sources the answers cite instead.

    aeosome prompts --missing-brand true

Two ways in

Create a read-only key in Settings, Integrations, then connect either way. Both read the same data.

  • MCP server

    For Claude, Cursor and any assistant that speaks MCP. The assistant picks the right tool for your question on its own.

    Claude Code
    claude mcp add --transport http aeosome https://api.aeosome.com/mcp \
  --header "Authorization: Bearer aeo_your_key"
    Cursor and clients that take a URL
    {
  "mcpServers": {
    "aeosome": {
      "url": "https://api.aeosome.com/mcp",
      "headers": {
        "Authorization": "Bearer aeo_your_key"
      }
    }
  }
}
    Set up your MCP client

  • Command line

    For terminals, scripts and scheduled jobs. Tables for people, JSON for programs with --json, and a fixed exit code for each kind of failure.

    Terminal
    npx @aeosome/cli login
npx @aeosome/cli dashboard
npx @aeosome/cli prompts --missing-brand
    Read the command line guide

What your agent can read

The same data as your dashboard. Each is a CLI command and an MCP tool.

  • whoami

    The workspace and plan this API key belongs to.

  • brief

    One-paragraph status of the store's AI visibility this week: what slipped, which rival is named instead, and what changed.

  • dashboard

    Headline AI visibility KPIs with trend against the prior period: visibility, top-3 presence, share of voice, citation health, shopping appearance.

  • share-of-voice

    Share of brand mentions in AI answers against competitors, optionally for one engine.

  • products

    Tracked products with their store URLs.

  • prompts

    Shopper questions sent to AI engines, whether the store was named, and which engines showed it. Use a question id with get_prompt_responses.

  • responses

    The full AI answers to one shopper question, per engine, with the products and sources each answer cited.

  • competitors

    Competitors ranked by share of voice in AI answers, with prompts won and citation share.

  • citations

    The domains AI engines cite most in answers about the store's products.

  • opportunities

    Sites that AI engines cite for competitors but never for the store: the outreach list.

  • fixes

    Fixes to raise AI visibility, with priority, expected impact and step-by-step implementation guides.

  • progress

    Open issues by category, fixes applied in the last 7 and 30 days, and the measured visibility change after fixes.

  • alerts

    Recent alerts: visibility drops, new competitors, and findings.

  • usage

    Plan limits and current usage: products, answers this month, team seats and the billing period end.

Read-only, and yours to revoke

A key can read one workspace and change nothing in it. Revoke it in Settings and it stops working at once. API access comes with every paid plan.

Compare plans or read how API keys work

Questions about the MCP server and CLI

Find out which AI answers name your store

Connect your store. We ask AI assistants the questions your shoppers ask, then show which answers name you, which name a competitor, and what to fix first.

Get started for Free