Claude-ready website intelligence through one Remote MCP endpoint

Add https://api.santosautomation.com/mcp to Claude once and seven website-intelligence tools appear in every conversation. This is the same production MCP server the Grok integration uses — one endpoint, one registry listing, different client setup. No API key, no server to run, no OAuth.

One difference worth knowing up front: Claude registers the server once, Grok names it per call. A Grok request carries the endpoint inline in its owntools array, so there is nothing to install. Claude keeps a connector on your account (or in your Claude Code config), so it must be added before the tools exist in a conversation — and toggled on for that chat.

Live MCP endpoint Grok setup API docs

Step one

Add the connector

Claude.ai and Claude Code are configured separately. A connector added in Claude.ai settings is not visible to Claude Code, and vice versa — adding it in one place and calling it from the other is the most common reason the tools appear missing. Pick the surface you actually use.

Claude.ai

Open Settings → Connectors → Add custom connector, paste the endpoint below, and connect. Claude handles discovery from there — it calls tools/list, reads the input schemas, and decides when to invoke. Custom connectors require a paid Claude plan. If the server does not appear in your connector list afterwards, it did not save; add it again.

https://api.santosautomation.com/mcp

Claude Code

Register it from the CLI. Use -s user — the default local scope registers the server for the current directory only, so it vanishes the moment you open another project.

claude mcp add -s user --transport http santos https://api.santosautomation.com/mcp

# confirm it registered
claude mcp list
# santos: https://api.santosautomation.com/mcp (HTTP) - ✔ Connected

Restart the session afterwards — MCP servers are read at startup, so the tools are not callable until Claude Code reloads. If claude mcp list shows entries prefixed claude.ai, those are your Claude.ai connectors mirrored into the CLI; a Santos entry missing from that list means the Claude.ai connector was never saved.

Either way

Nothing else to configure. The endpoint is public and unauthenticated — there is no OAuth step, no token to paste, and no key to manage. If a client prompts for OAuth it is guessing; Santos requires none. Verify the exact tool list Claude will see from your own shell:

curl -X POST https://api.santosautomation.com/mcp \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Step two

Ask Claude to use it

Paste this once the connector is attached. It audits a real page, so you can check the answer against what the page actually says.

Use the Santos tools to audit https://www.santosautomation.com/integrations/claude. Return the Website Intelligence score, the four dimension scores, and the three highest-impact fixes.

Claude picks audit_website_preview — a free tool, so no wallet is involved — and returns the Website Intelligence score, the four dimension scores, and prioritized fixes. If that comes back, the connector is working.

Reference

What Claude sees

Seven tools. Three run directly and return a result inline. Four validate the target and hand back a payment request — see the next section for what that means.

Tool	Cost	Behavior
`audit_website_preview`	Free	Runs a Quick Intelligence Audit and returns the report inline
`extract_page_markdown`	Free	Returns one page as clean Markdown with title, links, and word count
`extract_structured_data`	Free	Returns JSON extracted against your own JSON Schema, re-validated before return
`audit_agent_readiness`	$0.075	Returns the x402 handoff for the full Agent Readiness audit
`feed_parse`	$0.003	Returns the x402 handoff for RSS / Atom / JSON Feed normalization
`link_map`	$0.003	Returns the x402 handoff for a categorized link map
`summarize`	$0.033	Returns the x402 handoff for a structured page summary

Free tools share one quota: one call per day per identity. Without a token that identity is the calling IP; pass the optional token argument (a verified-email token from /api/leads/verify/*) and the quota moves onto that user instead.

Free tools run directly

The three preview tools execute end to end inside the connector call and return the result to Claude. No wallet, no payment, no configuration beyond attaching the endpoint. This is the whole experience for evaluation and for most conversational use.

Paid tools hand back a request

A paid tool does not charge anything and does not ask Claude for credentials. It validates the target and returns a canonical x402 HTTP request — the price, the network, and the exact URL to pay for. Claude does not sign x402 payments and does not manage a wallet. Completing one requires a wallet-enabled application or an agent wrapper holding its own key. See the handoff pattern for a worked example.

Verification

One server, one listing

Claude and Grok connect to the same endpoint. There is no Claude-specific server, no duplicate registry entry, and no separate tool set — only a different client setup. The listing is published in the official MCP Registry under com.santosautomation/site-audit and its streamable-http remote is exactly the endpoint above.

Official MCP Registrycom.santosautomation/site-audit — verify the remote URL MCP endpointStreamable HTTP · seven tools · strict schemas OpenAPI 3.1Typed operations, schemas, errors, and x402 behavior Capability manifestSelection guidance, pricing, limits, and support llms.txtLow-noise service guide and canonical machine links StatusLive availability of every capability

Questions or a problem connecting: info@santosautomation.com.