What is MCP?#
The Model Context Protocol (MCP) lets AI assistants like Claude, ChatGPT, and Cursor call external tools over a standard API. Reckon’s MCP server exposes email verification and credit balance as MCP tools, so you can verify addresses and check credits from inside your AI workflow without leaving the chat. Endpoint: https://mcp.reckonapp.io/mcp (Streamable HTTP).
Authentication#
When you add the Reckon MCP server in Claude (or another client that supports OAuth), you'll be prompted to sign in with your Reckon account (email and password). You need a Reckon account; if you don't have one, you can do that here.After you allow access, a dedicated API key is created in your Reckon account for that client so that usage can be tracked/reported per connection; you can view, pause, or revoke it anytime under API Keys. The client then uses that session for all tool calls (verify email, check credits) against your account.
Rate limits#
Rate limits apply per account. When you exceed the limit, the server returns a rate-limit error and may include a Retry-After value (seconds) so clients can back off and retry.
Quick setup#
Reckon MCP endpoint: https://mcp.reckonapp.io/mcp (Streamable HTTP).Add the endpoint above in your client's MCP settings. When you connect, you'll be prompted to sign in with your Reckon account; after you allow access, tools use that session.
verify_email#
Verifies an email address and returns deliverability status, format validity, domain info, and risk flags (disposable, role-based, accept-all, mailbox full, plus addressing). Each successful call consumes one Reckon credit.| Field | Type | Required | Description |
|---|
email | string | Yes | The email address to verify. |
{
"email": "user@example.com"
}
Example success response (summary of fields):{
"email_address": "user@example.com",
"status": "valid",
"status_meta": {
"valid_format": true,
"domain_exists": true,
"accept_all": false,
"disposable": false,
"role_based": false,
"mailbox_full": false,
"plus_address": false
},
"domain_meta": {
"domain": "example.com",
"mailbox_provider": "example.com"
},
"errors": {},
"performed_at": "2026-02-13T14:00:00.000Z",
"performed_in_ms": 3500,
"requestId": "..."
}
The MCP tool returns the full API response as JSON text content.
check_credits#
Returns the current email verification credit balance for the account associated with your API key. No input required.Input: None (empty object).Example success response:{
"balance": 1000,
"asOf": "2026-02-13T14:00:00Z"
}
The MCP tool returns the full API response as JSON text content.
Examples#
These examples show how the MCP server behaves when you use it from an AI assistant (e.g. Claude, ChatGPT, Cursor).Example 1: Verify an email address#
What happens:
The assistant calls the verify_email tool with the given address. The server sends the request to the Reckon API and returns the result. You see the full verification response: status (e.g. valid, invalid, unknown), format and domain checks, and risk flags (disposable, role-based, accept-all, etc.). One Reckon credit is consumed for a successful verification.
Example 2: Check credit balance#
User prompt:
“How many verification credits do I have?” or “What’s my Reckon credit balance?”What happens:
The assistant calls the check_credits tool (no input required). The server returns the current balance for the account tied to your API key. You see a short summary, e.g. “You have 1,000 credits” (from the balance and asOf fields in the response). No credits are consumed.
Example 3: Verify then check credits (or handle an error)#
What happens:
The assistant first calls verify_email with sales@mycompany.com. You get the deliverability and risk details for that address, and one credit is used. Then it calls check_credits and reports your updated balance (e.g. “Verification complete. You have 999 credits remaining.”). If you have no credits, the first call fails with “Insufficient credits” and the assistant can explain that you need to add credits at app.reckonapp.io; the assistant may still call check_credits to confirm the balance is zero.
Error handling#
Errors are returned as MCP/JSON-RPC errors with messages suitable for display. Underlying HTTP status from the Reckon API is reflected as follows:| Condition | Message / behavior |
|---|
| 400 — Invalid request | “Invalid request parameters.” Check input (e.g. valid email format for verify_email). |
| 401 — Not authenticated | Sign in again or re-authorize the MCP server in your client. |
| 402 — Insufficient credits | “Insufficient credits. Visit https://app.reckonapp.io to add credits to your account.” Add credits in the Reckon dashboard. |
| 429 — Rate limit | “Rate limit exceeded.” The error may include a retry-after value; wait before retrying. |
| 5xx — Server error | “Internal server error (status). Please try again.” Retry later or contact support if it persists. |
| Other 4xx | “Request failed (status). Please try again.” Check request format and parameters. |
Troubleshooting#
"Insufficient credits" (402)
Your account has no credits left. Add credits at app.reckonapp.io. “Rate limit exceeded” (429)
You've hit the rate limit. Wait before retrying; the error may include a suggested wait time.
“Internal server error” or “Request failed (404)”
For 5xx, the Reckon API is temporarily failing; try again later. For 404, the request path or upstream API may have changed; check the docs or contact support.
OAuth: "Missing or invalid access token"
The client must send Authorization: Bearer <access_token> on every request. If your client doesn't (e.g. some versions of MCP Inspector), you can get a token at mcp.reckonapp.io/oauth/debug-token and use the MCP URL with ?access_token=... (or the X-Access-Token header).
Support#
For help with the MCP server or billing:
Privacy policy#
Modified at 2026-02-16 21:33:22