Rate limits

The Customer API allows 120 requests per minute. That’s the production setting today, and it’s generous enough that most integrations never notice it. The window slides. It isn’t a bucket that resets on the minute — the server counts requests made in the last 60 seconds, continuously.

What counts toward the limit?

One shared budget covers three surfaces:

every /api/public/v1/* route
the MCP server at /api/mcp
both POST agent-ingest endpoints

So an assistant querying through MCP and a nightly export job spend from the same 120.

The limit is counted per client IP address, not per API key. Two services sharing one NAT gateway or one Kubernetes egress IP share one budget, even with separate keys. Creating another key does not buy you more throughput.

What happens when you go over?

You get 429 Too Many Requests with this body:

{ "error": "rate limit exceeded, try again later" }

and one header:

Retry-After: 12

Retry-After is a whole number of seconds, at least 1 and never more than 60. It’s the real wait — the time until your oldest request ages out of the window, not a fixed penalty. Sleep for that long and your next call goes through.

Are there quota headers?

Retry-After on a 429 is the rate-limit signal the API sends. Successful responses don’t carry X-RateLimit-Limit, X-RateLimit-Remaining or X-RateLimit-Reset, so track your own call count if you need to know where you stand before you hit the limit.

How should you handle it?

Retry on 429, and honour Retry-After rather than guessing.

import time, requests

def get(url, key, attempts=4):
    for i in range(attempts):
        r = requests.get(url, headers={"Authorization": f"Bearer {key}"})
        if r.status_code != 429:
            r.raise_for_status()
            return r.json()
        time.sleep(int(r.headers.get("Retry-After", 5)))
    raise RuntimeError("still rate limited after retries")

Two habits keep you well clear of the ceiling:

Widen the range instead of looping over days. One call with range=90d beats 90 calls with a date filter. See Filters and time ranges.
Prefer the endpoints that answer in one request. /industry/metrics returns the ranking, the time series and per-competitor sentiment together.

A daily sync of a five-brand workspace is roughly 30 requests. You’d have to run it every 15 seconds to trouble the limit.

Do you need more than 120?

Email nitish@fixaeo.com with the shape of your workload. The limit is a server setting, not something baked into the build, so it can be raised for an account that needs it.

Authentication

Keys, rotation, and what a 401 means.

Filters and time ranges

Fetch wider windows in fewer calls.

Customer API

The full error table and what each status means.

MCP server

Shares this budget — worth knowing if you run both.

Getting started

Public API

What counts toward the limit?

What happens when you go over?

Are there quota headers?

How should you handle it?

Do you need more than 120?

Authentication

Filters and time ranges

Customer API

MCP server

​What counts toward the limit?

​What happens when you go over?

​Are there quota headers?

​How should you handle it?

​Do you need more than 120?

​Related

Authentication

Filters and time ranges

Customer API

MCP server

What counts toward the limit?

What happens when you go over?

Are there quota headers?

How should you handle it?

Do you need more than 120?

Related