Free score

For merchants who work in an AI assistant

Run Beaconed from the assistant you already talk to.

Add Beaconed's hosted MCP server to your assistant with one address and a sign-in. Ask it to score your catalog or draft a listing, and it does the work through your account. The approval rules do not change: it drafts, you publish.

Connect your assistant Setup instructions

No install and no API key. The npm MCP server and command line tool are still available below.

prompt · paste into your assistant

Read https://beaconed.ai/agent-setup and follow ALL the instructions to connect yourself to Beaconed so you can help optimize my store's product listings. Ask me before making any config changes.

Hosted, no install

Connect with one address.

Beaconed runs the MCP server for you. Add it to any assistant that supports remote MCP connectors, sign in, and you are done.

server address

https://mcp.beaconed.ai/mcp

1. Add the connector

Paste the address into your assistant

In Claude, open Settings, then Connectors, and choose Add custom connector. Give it a name and paste the address. Other tools that support remote MCP connectors take the same address.

2. Sign in and allow

Approve it on Beaconed

Your assistant opens a Beaconed sign-in. Pick the account to connect (one you own or administer), check what the connection is allowed to do, and choose Allow access. Beaconed shows the tool by name and marks it unverified, which is expected.

3. Ask

Talk to it like you would a colleague

No API key and nothing to install. Access lasts 90 days, then your assistant asks you to sign in again.

Things to ask once it is connected

  • “Which ten products have the lowest readiness scores, and what is missing on each?”
  • “What is waiting in my review queue right now?”
  • “Why does my best seller score lower than the rest of the catalog?”
  • “Draft new descriptions for those ten products.” needs write access
  • “Approve the three drafts I liked and publish them.” needs write access

Reading scores and the review queue always works. Drafting, approving and publishing work only when the consent screen listed those permissions and you allowed them. Drafts use your credits, and publishing still waits for you to say so.

See or revoke access

Every connection is listed under Connected assistants (beaconed.ai/mcp/connections), with a Disconnect button for each one. The page opens on your default account.

What this means if you are not a developer.

You type a sentence. The assistant uses your Beaconed account to do the work and reports back. Nothing reaches your store without you publishing it.

Ask for it

“Score my catalog and show me the worst ten”

The assistant reads your scores through Beaconed and gives you the list, with what is missing on each product.

Get drafts

“Draft descriptions for those ten in my voice”

Beaconed uses your voice profile and each product’s own facts. Drafts appear in your review queue, not on your store.

Keep control

“Publish the three I approved”

Publishing needs your approval in the app and in a chat window alike. Anything published can be reverted.

Install it yourself

Prefer to run it locally?

Both packages are on npm and use an API key from your account settings. Use them for editors without remote connectors, scripts, and scheduled jobs.

@beaconed/mcp

The MCP server

Model Context Protocol is how an assistant reaches tools outside itself. Register the Beaconed server once and your assistant can list products, read scores, request drafts and read the review queue, using your API key.

bash · install 
npm install -g @beaconed/mcp
npm install -g @beaconed/cli

beaconed auth login
  • Works with Claude Code, Claude Desktop, Cursor, Cline and Zed
  • Scoped to the account the key belongs to
  • Cannot publish without a person approving

Register the server

@beaconed/cli

The command line tool

For scripts, scheduled jobs and anyone who would rather type than click. The same operations as the API, one command each.

bash 
beaconed products list --max-score 60
beaconed products optimize prod_8Kq2 --fields description,alt_text
beaconed optimizations list --status pending

Full command reference

Create an account, then connect your assistant.

Create an account, add the connector address to your assistant, and sign in. If you would rather see the product first, run the free score on your store.

Create an account, free Run the free score

Writing the setup yourself?

The agent setup page is written for an assistant to read and follow, with config for each editor.

Questions: info@beaconed.ai.