Connect your agent to Verithia

Verithia speaks the Model Context Protocol. Any MCP-capable client (Claude Code, Cursor, Claude Desktop, or your own code) connects to a single HTTPS endpoint and gets every tool your tier includes. There is nothing to install and nothing to run locally.

The two things you need

Endpoint   https://verithia.com/mcp
Transport  HTTP (JSON-RPC 2.0 over POST)
Auth       Authorization: Bearer YOUR_KEY

That is the whole contract. Every plan sees all 27 tools and every engine; plans differ only by how many AI queries (credits) you get, not by which tools unlock. Everything below is just how to hand those two facts to a specific client.

Do not have a key yet? Request one here. Keys look like vk_....

Claude Code

One command in your terminal:

claude mcp add --transport http verithia https://verithia.com/mcp \
  --header "Authorization: Bearer YOUR_KEY"

Then in a session, /mcp lists the connected server and its tools. Your agent can now call citation_domains, serp_profile, page_audit, and the rest by name.

Cursor

Add Verithia to ~/.cursor/mcp.json (global) or .cursor/mcp.json (per project):

{
  "mcpServers": {
    "verithia": {
      "url": "https://verithia.com/mcp",
      "headers": { "Authorization": "Bearer YOUR_KEY" }
    }
  }
}

Reload Cursor. Verithia appears under Settings, MCP, with a green dot when it connects.

Claude Desktop

Claude Desktop reaches a remote MCP through the mcp-remote bridge. Edit your claude_desktop_config.json (Settings, Developer, Edit Config):

{
  "mcpServers": {
    "verithia": {
      "command": "npx",
      "args": [
        "-y", "mcp-remote", "https://verithia.com/mcp",
        "--header", "Authorization: Bearer YOUR_KEY"
      ]
    }
  }
}

Restart Claude Desktop. The tools show up under the connectors icon in the composer. (npx needs Node.js installed.)

Raw HTTP

Every tool is reachable with a plain POST. First, list what your key can call:

curl -s https://verithia.com/mcp \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

The response carries each tool's inputSchema, so you always know the exact arguments. Then call one. This asks which domains Gemini cites for a buyer query:

curl -s https://verithia.com/mcp \
  -H "Authorization: Bearer YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
        "jsonrpc":"2.0","id":2,"method":"tools/call",
        "params":{
          "name":"citation_domains",
          "arguments":{"query":"best crm for startups","engine":"gemini"}
        }
      }'

The result comes back as JSON text in result.content[0].text. The supported methods are the MCP standard set: initialize, tools/list, tools/call, ping.

Plans

Every plan gets all 27 tools and every engine. They differ only by how many AI queries (credits) you get per month. One credit = one grounded AI query; page audits, SERP profiles and rank checks are free.

Plan	AI queries / mo	Rate limit	Price
Starter	300	30 calls/min	$29/mo
Pro	1,500	120 calls/min	$99/mo
Agency	10,000	higher	$499/mo

Out of credits returns JSON-RPC -32003 (HTTP 402); a call past the per-minute rate limit returns -32002. GET /usage (with your key) shows your remaining balance. See the full tool list and pricing.

Reading the numbers honestly

Grounded AI is stochastic: ask the same question twice and about half the cited sources change. So the tools sample a distribution, they do not return a single fixed truth. Run a query a few times and act on what recurs, not on one answer. We measured exactly how far this goes, including against the live Gemini app and Google's AI Overview, in the fidelity study. Verithia measures. Your agent interprets.

Stuck? Email the address on your key request and I will help you connect.