Single verification
POST or GET /verify: request shapes and every response field.
Two equivalent endpoints verify one address:
# POST with JSON body
curl -X POST "https://api.kavik.email/api/v1/verify" \
-H "X-API-Key: eva_your_key" \
-H "Content-Type: application/json" \
-d '{"email": "jane@acmecorp.com"}'
# GET with query parameter
curl "https://api.kavik.email/api/v1/verify?email=jane@acmecorp.com" \
-H "X-API-Key: eva_your_key"Repeat lookups for the same address may be served from a short-lived cache for speed; each call counts toward your usage once per call.
With a test key, use the test addresses in the sandbox, such as safe@example.com. Anything else is refused with LIVE_KEY_REQUIRED.
Response fields
| Field | Type | Meaning |
|---|---|---|
email |
string | The address as submitted |
score |
integer 0-100 | Composite deliverability score |
risk |
enum | safe (70-100), risky (40-69), catchall (40-69: the domain accepts every address, so the mailbox itself is unconfirmed; never billed), invalid (0-39), unknown (no definitive answer; never billed) |
syntax_valid |
boolean | RFC-correct address syntax |
mx_found |
boolean | The domain publishes MX records. false with smtp_check: undeliverable means no mail server answers anywhere RFC 5321 falls back to |
smtp_check |
enum | deliverable, undeliverable, risky, unknown |
is_disposable |
boolean | Known disposable provider |
is_role_account |
boolean | Role address (info@, sales@, …) |
is_catch_all |
boolean | Domain accepts every local part; the address gets the catchall verdict and the score orders it inside a 40-69 band unless the mailbox was actually refused |
is_free_provider |
boolean | Free mailbox provider (Gmail, Outlook, …) |
spam_trap_risk |
enum | low, medium, high |
suggested_correction |
string or null | Did-you-mean suggestion when a domain typo is detected. Suggestion only: we never verify the corrected address on your behalf |
mx_records |
array | The domain’s MX hosts |
is_security_gateway |
boolean | Mail routes through Mimecast, Proofpoint, etc. |
security_gateway |
string | Gateway name or empty |
has_gravatar |
boolean | A Gravatar profile exists for the address |
gravatar_url |
string | Present when has_gravatar is true |
livemode |
boolean | true for a live key, false for a test key |
Batch verification
For up to 100 addresses in one call, use POST /api/v1/verify/batch with {"emails": ["a@x.com", "b@y.com"]}. Duplicates are removed before anything is charged. The response is {"total", "results", "livemode"}, and each result carries livemode too.
A batch answers within about 80 seconds. An address whose check is still running then, because its mail server is slow to answer or its provider takes checks a few at a time, comes back in a checking list instead of results. It is not billed in that call, and its check keeps running. Call the endpoint again with those addresses to get their results: the second call waits for the running checks instead of starting new ones. total counts the results, and the X-Checking-Count header gives the number under checking. Match results to your list by email, not by position.
Account
GET /api/v1/account returns the plan and checks left for the key you call it with. A live key and a test key both work.
curl "https://api.kavik.email/api/v1/account" \
-H "X-API-Key: eva_your_key"{
"data": {
"plan": "starter",
"livemode": true,
"period_start": "2026-11-01",
"period_end": "2026-12-01",
"allowance_renews": true,
"monthly_limit": 20000,
"total_used": 1250,
"remaining": 18750,
"pool_credits": 0,
"daily_used": 0,
"free_allowance_held": false,
"checks_per_minute": 20
}
}| Field | Meaning |
|---|---|
plan |
The plan of the account the key is charged to |
livemode |
true for a live key, false for a test key |
period_start, period_end |
The current period. period_end is empty on the Free plan |
allowance_renews |
false on the Free plan, whose 300 checks do not renew |
monthly_limit, total_used |
The period’s allowance and how much of it is used |
remaining |
Plan checks left plus credits. 0 for a Free account without credits, because its 300 checks are for the dashboard |
pool_credits |
Credit-pack checks held |
daily_limit, daily_used |
The Free plan’s daily cap, when one applies, and today’s use |
free_allowance_held |
The free allowance is withheld until a plan or a credit pack is bought |
checks_per_minute |
The account’s per-minute check budget |
sandbox |
Test keys only: requests_per_minute, max_bulk_rows, max_running_jobs, max_jobs_per_hour |
The answer is for the pool the calling key is charged to. The call is never charged, and it is limited to 60 requests a minute per account. Its X-RateLimit-Limit: 60 is the limit of this endpoint, not checks_per_minute.