The Hive Inspect API is a work in progress. It is available to a small group of accounts today, and the release for everyone is coming very soon. To ask for early access, message us from the chat bubble in your dashboard.
Rate limit
You can make 120 requests per minute. The limit is counted per key owner, so all the keys one person created, and the AI assistants they connected, share it. When you go over the limit, the API returns429. Wait a few seconds and try again. If your integration runs in a loop, add a growing delay between retries so it does not keep hitting the limit.
Monthly quota
Your company can make 10,000 requests per month. All of your company’s API keys share this allowance, and so do AI assistants, whether they use an API key or are connected with Sign in with Hive.- Only successful responses (status
200to399) count. On the MCP address, each message in a request counts. - The month starts at your company’s first successful API call, not on the 1st of the calendar month.
- From request 10,001 to 15,000, requests still work, and every response carries a
Hive-Quota-Warningheader. - After 15,000 requests, the API returns
429until the next reset. Retrying does not help before then. - Webhooks do not count toward the quota.
A
429 without these headers means you hit the per-minute rate limit: wait and retry. A 429 with them means the monthly quota is used up.
Status codes
What an error looks like
Errors return JSON with adetail field that says what went wrong:
409 on a duplicate contact, detail includes the id of the contact that already has the email, so you can update it:
Which errors to retry
- Retry:
429from the per-minute limit and any5xxresponse, after a short wait. - Wait for the reset:
429from the monthly quota (it hasHive-Quota-*headers). - Do not retry as is:
400,401,403,404,409and422. These return the same error until you change the request.

