Rate limiting reference

Arcjet rate limiting lets you define rules which limit the number of requests a client can make over a period of time.

Configuration

Each rate limit is configured on an exact path with a set of client characteristics and algorithm specific options.

Fixed window rate limit options

Tracks the number of requests made by a client over a fixed time window. Options are explained in the Configuration documentation. For more details about how the algorithm works, see the fixed window algorithm description.

JavaScript / TypeScript

// Options for fixed window rate limit
// See https://docs.arcjet.com/rate-limiting/configuration
type FixedWindowRateLimitOptions = {
  // "LIVE" will block requests. "DRY_RUN" will log only
  mode?: "LIVE" | "DRY_RUN";
  // How the client is identified. See https://docs.arcjet.com/fingerprints
  characteristics?: string[];
  // Time window the rate limit applies to (e.g. "1h", "60s", or seconds as a number)
  window: string | number;
  // Maximum number of requests allowed in the time window
  max: number;
};

Python

# Signature for arcjet.fixed_window
# See https://docs.arcjet.com/rate-limiting/configuration
def fixed_window(
    *,
    # Required. Mode.LIVE blocks requests; Mode.DRY_RUN logs only.
    mode: Mode,
    # Time window in seconds the rate limit applies to (integer only — Python
    # does not accept the "1h"/"60s" duration strings that the JS SDK does).
    window: int,
    # Maximum number of requests allowed in the time window
    max: int,
    # How the client is identified. See https://docs.arcjet.com/fingerprints
    characteristics: Sequence[str] = (),
) -> FixedWindow: ...

Fixed window example

Guard rate limits use tokenBucket / TokenBucket. There is no fixedWindow or slidingWindow helper on @arcjet/guard. Set maxTokens to the burst you want and refillRate / intervalSeconds to the sustained allowance.

For a complete tool example, see the token bucket example on this page. For adapter-specific options, see the Vercel Eve agent guard.

Sliding window rate limit options

Tracks the number of requests made by a client over a sliding window so that the window moves with time. Options are explained in the Configuration documentation. For more details about how the algorithm works, see the sliding window algorithm description.

JavaScript / TypeScript

// Options for sliding window rate limit
// See https://docs.arcjet.com/rate-limiting/configuration
type SlidingWindowRateLimitOptions = {
  // "LIVE" will block requests. "DRY_RUN" will log only
  mode?: "LIVE" | "DRY_RUN";
  // How the client is identified. See https://docs.arcjet.com/fingerprints
  characteristics?: string[];
  // The time interval in seconds for the rate limit
  interval: number;
  // Maximum number of requests allowed over the time interval
  max: number;
};

Python

# Signature for arcjet.sliding_window
# See https://docs.arcjet.com/rate-limiting/configuration
def sliding_window(
    *,
    # Required. Mode.LIVE blocks requests; Mode.DRY_RUN logs only.
    mode: Mode,
    # The time interval in seconds for the rate limit
    interval: int,
    # Maximum number of requests allowed over the time interval
    max: int,
    # How the client is identified. See https://docs.arcjet.com/fingerprints
    characteristics: Sequence[str] = (),
) -> SlidingWindow: ...

Sliding window example

Guard rate limits use tokenBucket / TokenBucket. There is no fixedWindow or slidingWindow helper on @arcjet/guard. Set maxTokens to the burst you want and refillRate / intervalSeconds to the sustained allowance.

For a complete tool example, see the token bucket example on this page. For adapter-specific options, see the Vercel Eve agent guard.

Token bucket rate limit options

Based on a bucket filled with a specific number of tokens. Each request withdraws a token from the bucket and the bucket is refilled at a fixed rate. Once the bucket is empty, the client is blocked until the bucket refills. Options are explained in the Configuration documentation. For more details about how the algorithm works, see the token bucket algorithm description.

JavaScript / TypeScript

// Options for token bucket rate limit
// See https://docs.arcjet.com/rate-limiting/configuration
type TokenBucketRateLimitOptions = {
  // "LIVE" will block requests. "DRY_RUN" will log only
  mode?: "LIVE" | "DRY_RUN";
  // How the client is identified. See https://docs.arcjet.com/fingerprints
  characteristics?: string[];
  // Number of tokens to add to the bucket at each interval
  refillRate: number;
  // The interval in seconds to add tokens to the bucket
  interval: number;
  // The maximum number of tokens the bucket can hold
  capacity: number;
};

Python

# Signature for arcjet.token_bucket
# See https://docs.arcjet.com/rate-limiting/configuration
def token_bucket(
    *,
    # Required. Mode.LIVE blocks requests; Mode.DRY_RUN logs only.
    mode: Mode,
    # Number of tokens to add to the bucket at each interval
    refill_rate: int,
    # The interval in seconds to add tokens to the bucket
    interval: int,
    # The maximum number of tokens the bucket can hold
    capacity: int,
    # How the client is identified. See https://docs.arcjet.com/fingerprints
    characteristics: Sequence[str] = (),
) -> TokenBucket: ...

When using a token bucket rate limit, each request must specify the number of tokens it wishes to withdraw from the bucket. This is done by passing a requested property to the protect function.

Token bucket example

For how to specify the number of tokens to request, see the token bucket request example.

import { launchArcjet, tokenBucket } from "@arcjet/guard";
import { guardTool } from "@arcjet/guard/vercel-eve/v0";
import { defineTool } from "eve/tools";
import { z } from "zod";

const arcjet = launchArcjet({ key: process.env.ARCJET_KEY! });

const lookupLimit = tokenBucket({
  bucket: "lookups",
  refillRate: 5,
  intervalSeconds: 10,
  maxTokens: 10,
});

export default guardTool(
  arcjet,
  defineTool({
    description: "Look up an order by ID",
    inputSchema: z.object({ orderId: z.string() }),
    async execute(input) {
      return { orderId: input.orderId, status: "shipped" };
    },
  }),
  {
    action: "order.looked-up",
    rules: (input) => [lookupLimit({ key: input.orderId, requested: 5 })],
  },
);

The amount of tokens to deduct from the bucket is specified in the requested option as a positive integer when calling the Arcjet protect function.

JavaScript / TypeScript

// Deduct 50 tokens from the bucket
// The value for `requested` must be a positive integer
const decision = await aj.protect(req, { requested: 50 });

Python

# Deduct 50 tokens from the bucket
# The value for `requested` must be a positive integer
decision = await aj.protect(request, requested=50)  # use aj.protect(request, ...) for arcjet_sync

Note

The requested amount must be a positive integer.

Identify users

Rate limit rules use characteristics to identify the client and apply the limit across requests. The default is to use the client’s IP address. However, you can specify other characteristics such as a user ID or other metadata from your application.

Caution

Avoid passing personal information such as an email address as a custom characteristic. The value is hashed on Arcjet’s servers, but it’s still a best practice to avoid sending us sensitive information.

In this example we define a rate limit rule that applies to a specific user ID. The custom characteristic is userId with the value passed as a prop on the protect function. You can use any string for the characteristic name and any string, number or boolean for the value.

import { tokenBucket } from "@arcjet/guard";

const lookupLimit = tokenBucket({
  bucket: "lookups",
  refillRate: 5,
  intervalSeconds: 10,
  maxTokens: 10,
});

const userId = "user123";
// Key the bucket on a trusted application identity, not a model argument.
const rules = [lookupLimit({ key: userId, requested: 5 })];

To identify users with different characteristics, such as IP address for anonymous users and a user ID for logged in users, you can use withRule (JS) or with_rule() (Python) to create augmented clients that use different characteristics. See the example in the custom characteristics section.

Decision

Arcjet provides a single protect function that is used to execute your protection rules. This requires a RequestEvent property which is the event context as passed to the request handler.

This function returns a Promise that resolves to an ArcjetDecision object. This contains the following properties:

• id (string) – The unique ID for the request. This can be used to look up the request in the Arcjet dashboard. It is prefixed with req_ for decisions involving the Arcjet cloud API. For decisions taken locally, the prefix is lreq_.
• conclusion (ArcjetConclusion) – The final conclusion based on evaluating each of the configured rules. If you wish to accept Arcjet’s recommended action based on the configured rules then you can use this property.
• reason (ArcjetReason) – An object containing more detailed information about the conclusion.
• results (ArcjetRuleResult[]) – An array of ArcjetRuleResult objects containing the results of each rule that was executed.
• ip (ArcjetIpDetails) – An object containing Arcjet’s analysis of the client IP address. For more information, see the SDK reference.

To check whether a rate limit rule returned a deny conclusion, use decision.isDenied() and decision.reason.isRateLimit() (JS) / decision.is_denied() and decision.reason_v2.type == "RATE_LIMIT" (Python).

You can iterate through the results and check whether a rate limit was applied:

JavaScript / TypeScript

for (const result of decision.results) {
  console.log("Rule Result", result);
}

Python

for result in decision.results:
    print("Rule Result", result)

This example logs the full result as well as each rate limit rule:

import { launchArcjet, detectPromptInjection } from "@arcjet/guard";

const arcjet = launchArcjet({ key: process.env.ARCJET_KEY! });
const inbound = detectPromptInjection();

export async function screenPrompt(prompt: string) {
  const decision = await arcjet.guard({
    label: "message.received",
    rules: [inbound(prompt)],
  });

  console.log("conclusion", decision.conclusion);
  console.log("reason", decision.reason);
  console.log("failed open", decision.hasFailedOpen());

  for (const result of decision.results) {
    console.log("rule result", result);
  }
}

Token bucket request

When using a token bucket rule, pass an additional requested prop (a positive integer) to the protect function. This is the number of tokens the client is requesting to withdraw from the bucket.

import { launchArcjet, tokenBucket } from "@arcjet/guard";
import { guardTool } from "@arcjet/guard/vercel-eve/v0";
import { defineTool } from "eve/tools";
import { z } from "zod";

const arcjet = launchArcjet({ key: process.env.ARCJET_KEY! });

const lookupLimit = tokenBucket({
  bucket: "lookups",
  refillRate: 5,
  intervalSeconds: 10,
  maxTokens: 10,
});

export default guardTool(
  arcjet,
  defineTool({
    description: "Look up an order by ID",
    inputSchema: z.object({ orderId: z.string() }),
    async execute(input) {
      return { orderId: input.orderId, status: "shipped" };
    },
  }),
  {
    action: "order.looked-up",
    rules: (input) => [
      // Deduct 50 tokens from the bucket.
      // The value for `requested` must be a positive integer.
      lookupLimit({ key: input.orderId, requested: 50 }),
    ],
  },
);

Rate limit headers

With a rate limit rule enabled, you can access additional metadata in every Arcjet decision result:

• max (number): The configured maximum number of requests applied to this request.
• remaining (number): The number of requests remaining before max is reached within the window.
• window (number): The total amount of seconds in which requests are counted.
• reset (number): The remaining amount of seconds in the window.

These can be used to return RateLimit HTTP headers (draft RFC) to offer the client more detail.

In JavaScript, use setRateLimitHeaders from @arcjet/decorate. In Python, use set_rate_limit_headers from the arcjet package. Both write RateLimit and RateLimit-Policy headers from the decision.

When several rate limit results are present, the tightest remaining budget is advertised. If two policies share the same max, no headers are written.

This would result in draft RFC response headers similar to the following:

...
< RateLimit: limit=10, remaining=5, reset=9
< RateLimit-Policy: 10;w=10
...

Error handling

Arcjet is designed to fail open so that a service issue or misconfiguration does not block all requests. The SDK also times out and fails open after 2000 ms by default. However, in most cases, the response time is less than 20 ms to 30 ms.

If there is an error condition when processing the rule, Arcjet returns an ERROR result for that rule and you can check the message property on the rule’s error result for more information.

If all other rules that were run returned an ALLOW result, then the final Arcjet conclusion is ERROR.

import { launchArcjet, tokenBucket } from "@arcjet/guard";
import { guardTool } from "@arcjet/guard/vercel-eve/v0";

const arcjet = launchArcjet({ key: process.env.ARCJET_KEY! });

const lookupLimit = tokenBucket({
  bucket: "lookups",
  refillRate: 5,
  intervalSeconds: 10,
  maxTokens: 10,
});

// Helpers default to onGuardError: "deny". The tool does not run if Guard
// cannot be evaluated. Set "allow" only when executing without a complete
// security decision is acceptable.
export const policy = {
  action: "order.looked-up",
  onGuardError: "deny",
  rules: () => [lookupLimit({ key: "user123", requested: 5 })],
};

Testing

Arcjet runs the same in any environment, including locally and in CI. You can use the mode set to DRY_RUN to log the results of rule execution without blocking any requests.

We have an example test framework you can use to automatically test your rules. Arcjet can also be triggered based using a sample of your traffic.

For details, see the Testing section of the docs.

Discussion