human forward · connector

Ask your AI assistant how findable you are.

Human Forward measures how findable a brand is in Google, in AI answers and to AI shopping agents, and gives it a list of fixes. This connector puts all of that inside ChatGPT, Claude, Meta's Muse or any assistant that speaks MCP.

You need a Human Forward account first: the Shopify app, or an account at app.humanforward.io for any other site. Settings shows a connect code. The assistant asks for your site and that code, nothing else.

Muse

Connectors, find Human Forward, Connect. Or ask Muse to add a custom connector at the MCP address below.

ChatGPT

Apps, find Human Forward, Connect. Or Settings, Apps and connectors, Advanced, Developer mode, Create, paste the MCP address, authentication OAuth.

Claude

Settings, Connectors, find Human Forward, Connect. Or Add custom connector and paste the MCP address.

Gemini

gemini.google.com, Settings, Connected apps, Add a custom app, paste the MCP address. Then type @ in a prompt and pick Human Forward. Personal accounts, English, US at the moment.

In ChatGPT the answers come back as cards: the score panel, the list with tick boxes and copy buttons, the engine answers by tab, agent readiness, and a live view while a check runs.

How findable is my store?
What should I fix first, and give me the prompt to do it.
What did Perplexity say about us in France?
Which questions do shoppers ask that we are not the answer to?
Can Muse buy from my site? What is missing?
Run the check again and tell me what changed.

mcp endpoint, streamable httphttps://app.humanforward.io/mcp

authentication

OAuth 2.1 with PKCE (S256). Dynamic client registration is on, so a client registers itself, sends the user to the authorise page, and the user proves the site is theirs with the connect code. Access tokens last 30 days, refresh tokens rotate. Discovery:

https://app.humanforward.io/.well-known/oauth-authorization-server
https://app.humanforward.io/.well-known/oauth-protected-resource

scopes

findability:read reads scores and answers, findability:run starts runs. No customer or order data is ever exposed.

tools

get_overviewThe latest findability score: overall out of 100, the search and AI answer parts, agent readiness, sentiment, and a one line verdict per engine.

get_todoThe to do list, biggest win first: title, why it matters, effort, points, the steps, and a ready made prompt to do it.

mark_doneTick an item off the list, or untick it.

get_engine_answersWhat each AI engine actually said when asked the five customer questions, with the judge's notes. Filter by engine and market.

get_agent_readinessCan AI shopping agents (Muse, ChatGPT, Google via UCP) discover, read and buy from the site. Each check with its result and fix.

get_site_checksEvery technical check on the site: schema, crawlers, speed, content. Pass, partial or fail.

get_shopper_questionsThe twelve questions shoppers ask AI assistants in this category, whether the brand came up on each engine, and which rivals came up instead.

get_historyScore history across runs: overall, the parts, and per engine. For trends and what changed.

run_checkStart a fresh findability check. One to three minutes.

run_prompt_researchStart the shopper questions research. Two to three minutes.

The connector is part of Pro and Scale. A Free account can connect, and every tool answers with what Pro adds and the upgrade link.

api key instead

A per site API key from Settings also works: https://app.humanforward.io/mcp/<key>, no OAuth. Good for scripts, not for a directory listing.

The connector reads findability results that already exist in the merchant's account and can start a new run. It never sees customers, orders, or payment details, and it cannot change a site. A merchant can disconnect any assistant from Settings at any time, and rotating the connect code does not affect existing connections.

Privacy · Terms for apps · support hello@humanforward.io · Human Forward Ltd, London