Setup Guide

HTTP API

All platforms

Call MCP tools from any language, platform, or AI agent using standard HTTP requests. Works with Claude Code, Cursor, custom agents, webhooks, and any HTTP client.

1. Get your API key

Create an API key from your admin dashboard. Keys use the hk_ prefix.

2. Discover available servers

curl https://mcp.lthn.ai/servers.json \
  -H "Authorization: Bearer YOUR_API_KEY"

3. Call a tool

curl -X POST https://mcp.lthn.ai/tools/call \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "server": "openbrain",
    "tool": "brain_recall",
    "arguments": { "query": "authentication decisions" }
  }'

4. Read a resource

curl https://mcp.lthn.ai/resources/plans://all \
  -H "Authorization: Bearer YOUR_API_KEY"

Endpoints

View OpenAPI Spec →

GET /.well-known/mcp-servers.json Agent discovery

GET /servers List all servers

GET /servers/{id} Server details + tools

POST /tools/call Execute a tool

GET /resources/{uri} Read a resource

Code Examples

Python

import requests

resp = requests.post(
    "https://mcp.lthn.ai/tools/call",
    headers={"Authorization": "Bearer hk_your_key"},
    json={
        "server": "openbrain",
        "tool": "brain_recall",
        "arguments": {"query": "recent decisions"}
    }
)
print(resp.json())

JavaScript

const resp = await fetch("https://mcp.lthn.ai/tools/call", {
  method: "POST",
  headers: {
    "Authorization": "Bearer hk_your_key",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    server: "openbrain",
    tool: "brain_recall",
    arguments: { query: "recent decisions" },
  }),
});
const data = await resp.json();

Authentication

Authorization Header (Recommended)

Authorization: Bearer hk_abc123_your_key_here

X-API-Key Header

X-API-Key: hk_abc123_your_key_here

Server-scoped keys

API keys can be restricted to specific MCP servers. If you get a 403 error, check your key's server scopes in your admin dashboard.

Rate limiting

Requests are rate limited to 120 per minute. Rate limit headers (X-RateLimit-Limit, X-RateLimit-Remaining) are included in all responses.

Discovery

Agents discover available servers automatically via the well-known endpoint:

curl https://mcp.lthn.ai/.well-known/mcp-servers.json

Returns the server registry with capabilities and connection details. No authentication required for discovery.