@xemahq/grant-cache

In-process TTL cache for policy decisions

Overview

A small, in-process LRU + TTL cache that sits in front of a policy check and remembers each decision keyed by its execution context. On a miss or expiry it calls a caller-supplied filler, and it coalesces concurrent requests for the same key onto a single in-flight call so a burst of identical checks hits the backend once. It also exposes predicate-based invalidation so callers can evict entries when grants change. No network or persistence — one cache instance per process.

When to use it

• Use it to cut repeated round-trips to a policy backend for the same execution context within a single process.

Installation

pnpm add @xemahq/grant-cache

Usage

import { GrantCache } from '@xemahq/grant-cache';

const cache = new GrantCache({
  filler: (ctx) => checkPolicy(ctx), // called on miss / expiry
});

const decision = await cache.check(executionContext);

License

Apache-2.0 © Xema — xema.dev