Zero dependencies, fixed-window, queued rate limiter for Axios: set how many requests per interval should perform immediately, other will be delayed automatically.

npm install axios-rate-limit

import axios from 'axios';
import rateLimit from 'axios-rate-limit';

const http = rateLimit(axios.create(), {
  limits: [
    { maxRequests: 5, duration: '2s' },
    { maxRequests: 2, duration: '500ms' }
  ]
})
http.get('https://example.com/api/v1/users.json?page=1')
http.getQueue()

limits is the recommended format. It accepts an array of independent fixed windows, and a request is executed only when all windows allow it.

const http = rateLimit(axios.create(), {
  limits: [
    { maxRequests: 100, duration: '1m' },
    { maxRequests: 10, duration: '1s' }
  ]
})

Each limits[] entry:

• maxRequests (number, required, > 0): max requests per window.
• duration (string or number, required, > 0): window size. If a number is provided, it is interpreted as milliseconds (duration: 1500 means 1.5 seconds). Strings support ms, s, m, h (examples: '500ms', '2s', '1m').

• queue (optional): custom queue implementation. Must support push(item) and shift(), and either length or getLength(). Sync and async queues are supported.
• shouldCountRequest (optional): predicate (config, response) => boolean. If it returns false, the limiter refunds one occupied slot (useful for cached responses).
• rateLimiter (optional): lets you pass an existing limiter instance, so multiple axios clients can share one quota.

Returned axios instance methods:

• getQueue(): returns current queue instance.
• getMaxRPS(): returns first window RPS view (maxRequests / (durationInMs / 1000)), or 0 when limiter is not configured.
• setRateLimitOptions(options): updates limiter options at runtime.
• setMaxRPS(rps): shorthand runtime update equivalent to setRateLimitOptions({ maxRequests: rps, perMilliseconds: 1000 }).

Also available from module:

• rateLimit.getLimiter(options): creates a limiter instance that can be shared across axios clients.

Single-window constructor shape (maxRequests with one of perMilliseconds, duration, or maxRPS) is still supported for compatibility. For new code, prefer limits[] format.

• Single rate limit — API enforces one limit; use one window via limits.
• Multiple rate limits — API enforces several limits (e.g. per second and per minute); use multiple windows.
• Custom queue — Pass your own queue (e.g. to log when requests are added or removed).
• Retrying failed requests — Use with axios-retry to rate-limit and retry failed requests (see issue #24).
• Integration with axios-cache-adapter — Don't count cached responses toward the limit (see issue #43).
• Change RPS on the fly — Update setMaxRPS/setRateLimitOptions at runtime and speed up queued cancellation handling (see issue #48).
• Mocking in Jest — How to mock axios-rate-limit in Jest so tests do not hit the network (see issue #51).
• Shared limiter — Reuse one limiter instance across multiple axios clients that share the same API quota.

Consider using Axios built-in rate-limiting functionality.