INITE Atlas Owners

INITE MCP — AI visibility, from inside your agent

Ask your assistant whether the answer engines can see a website, and get a number back.

> Can ChatGPT actually find stripe.com, or is it invisible?
  analyze_site(url: "stripe.com")   → run_id: cm4x…  get_analysis(run_id: "cm4x…")     → Score: 78/100

The audit reads a live site the way an assistant does — the identity files an engine looks for (llms.txt, ai.json and the rest), the crawler policy, the schema graph, page health — and then asks Claude, ChatGPT, Gemini and Perplexity whether they name it. One score out of 100 across eight weighted sections.

It is the same audit that runs at inite.ai/en/analyze.

Works before you sign in

Three of the tools run entirely on your machine — a robots.txt fetched, a file probed for, a page parsed. No account, no allowance, no call home.

> Is anything blocking AI crawlers on stripe.com?
  check_ai_access(url: "stripe.com")
  Retrieval: all 7 answer-engine crawlers may fetch the site.  Training:  3 of 8 blocked — Meta-ExternalAgent, Bytespider, Amazonbot.

The rest is inite.ai's AI visibility server, and it answers signed out too, as a guest: check_identity, the fix plan (visibility_fixes), a competitor read the same way (compare_sites), how the reading moved (visibility_history), and analyze_site as the survey - the audit without its two paid steps, two a day per address.

> What should firecrawl.dev fix first so ChatGPT can use it?
  visibility_fixes(url: "firecrawl.dev")
  1. [critical] No structured data on the site (about +3.5 points)     Do: Add Organization (and WebSite) JSON-LD to every page.  2. [high] 7 retrieval crawlers are blocked on …

Visibility and nothing else. INITE's services and booking are ai.inite/inite, the catalogue of AI tools is ai.inite/atlas.

Signing in makes the calls your account's: its tier of the audit (the teaser on a free plan, the whole audit on a paid one) and your audits: my_reports, and the fix plan and history built from them.

Install

Claude Desktop, Cursor, or any client that launches a stdio server:

{  "mcpServers": {    "inite": {      "command": "npx",      "args": ["-y", "@inite/visibility"]    }  }}

Then sign in once:

npx @inite/visibility login

That opens a browser, you approve, and the token is stored at ~/.config/inite/mcp.json with owner-only permissions. Authorization code with PKCE over a loopback redirect, the flow RFC 8252 prescribes for a native app. Nothing is written to the repository and no secret ships in the package.

Already have a token (CI, a shared config, a container):

{  "mcpServers": {    "inite": {      "command": "npx",      "args": ["-y", "@inite/visibility"],      "env": { "INITE_TOKEN": "…" }    }  }}

INITE_TOKEN wins over the stored file.

Client speaks remote MCP? Skip this package. Point it straight at https://inite.ai/api/mcp/analyzer: every open tool answers without a token, and a member tool answers 401 with WWW-Authenticate naming the protected-resource metadata, which names the authorization server.

Tools

Local — no account:

Remote — open without an account, more with one:

Two tools rather than one for the audit, because it is asynchronous. A single tool that blocked for a minute would be torn down by most clients' timeouts.

Commands

inite-visibility            run as an MCP server over stdio (what a client does)inite-visibility login      sign in through the browserinite-visibility whoami     say whether a usable token is present

What this package is

Two halves.

The local checks are real work done here: fetching, parsing, and the robots rules applied properly — most-specific group wins, longest matching rule wins, Allow breaks a tie. They cost nobody anything because your machine does them.

The remote half defines no schemas of its own. It asks inite.ai what it offers and forwards calls there, so that tool list is whatever the service implements today. A local copy would be a second source of truth, and the first thing it would do is drift.

What stays on the server is what costs something or is ours: four answer engines asked whether they name a site, and the weights that turn everything into one number. The local tools report facts; analyze_site reports a score.

Account and allowance

Without an account every open tool answers, under a per-address allowance: the identity check, the fix plan, the comparison and the history, and the audit as a survey, two a day. An account makes the audit its plan's own, exactly as on the website: the teaser on a free plan, the full pipeline - four answer engines asked - on a paid one. Plans are here.

Environment

Namespace

Published to the MCP registry as ai.inite/inite-visibility, a namespace held by proving control of inite.ai — the public half of the key is served at /.well-known/mcp-registry-auth.

Named by the domain rather than the code host on purpose: a service whose whole job is being legible to machines should tell them who it belongs to in its own name.

A note on the command name

npx @inite/visibility resolves because there is exactly one binary in the package. npm looks for a command matching the package name with the scope stripped — visibility — does not find it, and runs the only one there is.

The binary is called inite-visibility rather than visibility on purpose: a scoped package has no business claiming a word that general in your PATH on a global install. The cost is that a second binary would break the line above for every client, so a test pins it at one.

If you are invoking it from a directory that contains this package's own package.json, npm prefers the local copy and finds no linked command. Name it explicitly there:

npx -p @inite/visibility inite-visibility login

Development

npm installnpm run buildnpm test

The tests cover what the bridge can get wrong without saying so: dropping the credential, turning "signed out" into "broken", swallowing an error into a success, and letting a stale file beat the environment.

Contribute with an AI agent

[Open in GitHub Codespaces]

You don't need to be an engineer. Open the repo in your coding agent and say /contribute (Claude Code) — or "read AGENTS.md and help me add …" in any other agent. AGENTS.md holds the rules, the commands CI runs and the idea → PR loop; CONTRIBUTING.md is the short version. Only want to suggest something? /spec-idea, or the Product idea form.

Licence

MIT. The service it talks to is INITE.