Foundbyai MCP Server

Are you found by AI? MCP server

Hosted MCP server for Are you found by AI?, the AI visibility tool built by Tech Horizon Labs. The product asks AI engines the questions a business's buyers actually type, records whether the business is named, who is named instead and which sources the engines read, and re-measures on a schedule. On Pro the customer chooses how often (weekly, monthly or whenever they like), and each check spends tokens from one balance shared across every site on the account; the free plan re-checks monthly. This server gives the business owner's own AI live access to that measurement mid-conversation.

Try it now, no account

In Claude Code, one line:

claude mcp add --transport http found-by-ai https://areyoufoundbyai.com/mcp/demo

Or add this URL to the Claude app (Settings, Connectors, Add custom connector), Cursor, or any other MCP client as a remote server (Streamable HTTP). No key and no signup:

https://areyoufoundbyai.com/mcp/demo

The demo is connected to our own live monitoring of areyoufoundbyai.com, so you are interrogating real production data. Ask your AI: "how visible is this business to AI, and what would you fix first?"

What your AI gets

The public demo lists twenty tools: nineteen read tools over live measurements, plus one action. They are scoped to the monitor behind the token, so none of them takes a URL. The evidence tools (crawler access, AI traffic, agent view, benchmark, share of voice, cited queries, source profile, citation sources, schema evidence, regional visibility) read saved measurements only, so calling them never starts a new scan or uses the scan allowance:

• get_visibility: AI Visibility and AI Readiness scores (each out of 100) with the previous week, the separate off-site Footprint score, and the subscores
• get_answers: the answer each engine gave to each tracked buyer question, with the competitors it named and the web searches it ran first
• get_question_trajectories: question-by-question history across the engines
• get_rivals / get_share_of_voice: who AI names instead of you, and how often
• get_cited_queries: the questions where a given competitor gets named
• get_mentions: new pages on the web that mention the business, when a mention check has been run
• get_fix_plan: the prioritised fix plan, in plain language
• get_citation_sources: the sources the engines actually cite
• get_source_profile: a profile of any cited source, including where a business gets listed on it
• get_post_brief: a content-drafting brief built from verified data
• get_ai_traffic: AI-referred visitors measured on the business's own site
• get_agent_view: what an AI agent's browser actually sees on the site
• get_crawler_access: which AI crawlers the site's robots.txt allows or blocks
• get_schema_evidence: the structured data in the site's HTML compared with what renders
• get_regional_visibility: naming results by area
• get_benchmark: the business against its category on our index
• get_personas: the buyer personas behind the tracked questions
• get_context: the full weekly pack in one call
• request_rescan: the one action, a fresh measurement capped by the plan's on-demand allowance (5 per rolling 7 days on a subscription that keeps its earlier terms). On a token account it queues nothing and asks for the token price to be confirmed in the workspace. On the demo it returns a worked example and queues nothing

On a customer's own token

A customer's own token lists more than the demo. The names and counts below come from the hosted service's source with its current release settings (token billing and the client workspace on); they have not been listed against a live customer token, so check your own tools/list.

Three more saved-only reads (twenty-three with the twenty above):

• get_balance: the site's existing on-demand scan allowance, queued status and cooldown. It does not show the token balance
• get_weekly_brief: the saved weekly brief
• get_grants: the saved grants source check

Seven workspace reads (thirty in total):

• get_token_balance: the account's shared token balance and expiry dates
• get_token_run: the saved state and result of one token run
• list_reports, get_report, list_deleted_reports, list_report_versions: saved client reports and their versions
• get_brief: the saved client brief

When the account owner turns on editing for the site's connection, nine workspace actions as well (thirty-nine in total). Each needs confirmed: true, recording that the person agreed:

• quote_token_run: a ten-minute price quote for a run, without spending
• confirm_token_run: after the person agrees to the quoted price, reserves the tokens and queues that run
• set_brief, save_report, update_report, delete_report, restore_report, restore_report_version, regenerate_report: save and manage the client brief and reports

There is no schedule tool. The customer chooses how often checks run in the signed-in workspace.

Pricing is on areyoufoundbyai.com/pricing: Pro is US$79 a month for 14,000 tokens, or US$869 a year, and a check of 25 buyer questions plus a fresh site read uses 3,150 tokens. Token packs (US$10 for 1,000 tokens, US$20 for 2,000, US$50 for 5,000) need no subscription.

Customers

Every monitored site has its own token, on the Free plan as well as Pro. The recommended form is the static endpoint with the token in a header, because URLs end up in server logs, proxies and browser history and headers do not:

curl -X POST https://areyoufoundbyai.com/mcp \
  -H "authorization: Bearer <your-token>" \
  -H "content-type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

x-api-key: <your-token> works too, and the path form https://areyoufoundbyai.com/mcp/<token> keeps working for anything already pointed at it. Agency keys get a separate network catalogue over the sites they are allowed to see. Setup guide, with a one-click Cursor link and the Claude Code command: areyoufoundbyai.com/guides/connect-your-ai

The readiness method and the audit suite are open source at techhorizonlabs/thl-open. The measurement engine itself is not.

Registry manifest

server.json is the manifest for the official MCP registry. Its version is the version of this registry entry, not of the server software: the registry needs a new version number each time the entry's metadata changes, so the entry is at 1.1.1 while the hosted server reports 1.1.0 in its serverInfo. Both point at the same hosted server.

Tech Horizon Labs, Noosa, Australia. hello@techhorizonlabs.com