@askalf/redisflex

One Redis API. Two modes. Same call sites.

Switch between real Redis (ioredis) and in-process Redis (Map + EventEmitter) with one line of config. Production runs on real Redis; dev / standalone / "no-Docker mode" runs in-process. Same get/set, counters, hashes, lists, sets, sorted sets, pub/sub — drop the Redis server when you don't need it.

Also ships a BullMQ-shaped in-memory queue so you can drop the Redis dep entirely for queueing too.

npm install @askalf/redisflex

Why

Most apps use Redis for three things: cache, pub/sub, and queueing. Most dev environments don't want to spin up a Redis server for any of them. Most CI environments REALLY don't want it.

redisflex swaps a real Redis connection for an in-process implementation that speaks the same surface — get/set, counters, hashes, lists, sets, sorted sets, pub/sub, expiry, and a sliding-window-rate-limit-shaped Lua eval. Plus a tiny BullMQ-API-compatible queue if you use BullMQ for jobs.

Your call sites stay identical. Flip the mode in config and your app no longer needs Redis to run.

Use it

Direct

import { createRedisAdapter } from '@askalf/redisflex';

// Production — real Redis
const redis = createRedisAdapter({
  mode: 'ioredis',
  url: process.env.REDIS_URL!,
});

// Dev / standalone — in-process
const redis = createRedisAdapter({ mode: 'memory' });

await redis.set('user:1', 'alice');
await redis.publish('events', 'user.created');

From environment

import { createRedisAdapterFromEnv } from '@askalf/redisflex';

// REDISFLEX_MODE=memory     → in-process
// otherwise                 → ioredis at $REDIS_URL (or redis://localhost:6379)
const redis = createRedisAdapterFromEnv();

Custom env-var names:

const redis = createRedisAdapterFromEnv({
  modeEnvVar: 'MYAPP_MODE',
  urlEnvVar: 'MYAPP_REDIS',
  defaultUrl: null, // null = throw if URL missing, instead of defaulting
});

In-memory queue

import { InMemoryQueue, InMemoryWorker } from '@askalf/redisflex';

const queue = new InMemoryQueue('emails');
const worker = new InMemoryWorker(
  'emails',
  async (job) => {
    await sendEmail(job.data);
  },
  { queue, concurrency: 4 },
);

worker.on('completed', (job) => console.log('sent', job.id));
worker.on('failed', (job, err) => console.error('failed', job.id, err));
worker.on('drained', () => console.log('all caught up')); // once per drain

await queue.add('welcome', { to: 'alice@example.com' });
await queue.add('reminder', { to: 'bob@example.com' }, {
  delay: 60_000,
  attempts: 3,
  backoff: { delay: 5000 }, // exponential: 5s, 10s, 20s
});

await queue.getJobCounts();
// { delayed: 1, waiting: 0, active: 1, completed: 0, failed: 0 }

await worker.close(); // waits for in-flight jobs; close(true) skips the wait

The shape matches BullMQ's Queue / Worker so you can swap to real BullMQ later by changing imports. Jobs move through the BullMQ states (delayed → waiting → active → completed/failed; a retrying job counts as delayed while its backoff timer is pending) — getJobs(states) and getJobCounts(...states) report them, and the queue re-emits completed/failed so you don't need a separate QueueEvents object. Completed/failed history is retained for introspection, capped at 1000 each (oldest dropped); set removeOnComplete/removeOnFail per job to skip retention.

What's covered

Family	Operations
Key/Value	get, set (incl. trailing EX <seconds> / PX <ms>), setex, mget, del, keys, exists
Counters	incr, decr, incrby, hincrby
Hashes	hset, hget, hgetall, hdel
Lists	lpush, rpush, lpop, rpop, llen, lrange
Sets	sadd, srem, smembers, sismember, scard
TTL	expire, pexpire, ttl, pttl
Sorted Sets	zadd, zcard, zscore, zrange, zrangebyscore, zrem, zremrangebyscore (incl. -inf/+inf and exclusive (n bounds)
Pub/Sub	publish, subscribe, unsubscribe, psubscribe, on('message'/'pmessage')
Scripting	eval (sliding-window-rate-limit shape recognized in memory mode)
Lifecycle	duplicate, quit

That's enough surface for cache, pub/sub, BullMQ-style queues, sliding-window rate limits, and most idiomatic Redis usage. Streams, geo, cluster, MULTI/EXEC, bitmap ops aren't covered. Open an issue if you need one — most are mechanical to add.

Note on set options: memory mode handles EX/PX and throws on anything else (NX, XX, KEEPTTL, ...) instead of silently ignoring it, so the two modes can't quietly diverge. ioredis mode passes all options through to the server.

Lua eval in memory mode

Memory mode recognizes the canonical sliding-window-rate-limit Lua script (ZADD + ZREMRANGEBYSCORE + ZCARD + EXPIRE, args layout [key, now, window-ms, member?]) and returns the post-add count. That's enough for typical rate-limiter use.

Anything else: returns 0 and prints a stderr warning once per process. Pass { silentEvalFallback: true } to suppress the warning. If you need a different Lua script handled natively, open an issue — they're ~5 lines each in memory-adapter.ts.

Pub/Sub: duplicate() semantics

Real Redis requires a separate connection for pub/sub vs commands. IoRedisAdapter.duplicate() calls Redis.prototype.duplicate() — a fresh TCP connection. MemoryRedisAdapter.duplicate() returns a new adapter that shares the underlying EventEmitter + state, so pub/sub fans out the same way real Redis does.

What it isn't

• Not a cluster client. ioredis mode supports cluster URLs; memory mode is single-node by definition.
• Not durable. Memory mode loses everything on process restart. Persistence is your application's problem (database run table, snapshot to disk, etc.).
• Not a full Lua interpreter. See above.
• Not a substitute for real Redis under load. The data structures are correct but not optimized for high throughput. Use real Redis in production.

License

MIT — see LICENSE.

Also by askalf

Project	What it does
arnie	Portable IT troubleshooting companion. Networking, AD, Windows Update, package managers, log triage, hardware checks.
browser-bridge	Stealth headless Chromium in a container. CDP on 9222 — Playwright/Puppeteer/MCP-compatible.
dario	Local LLM router. Use your Claude Max/Pro subscription as an API.
deepdive	Local research agent. Plan → search → fetch → extract → synthesize. Cited answers.
git-providers	Unified GitHub + GitLab + Bitbucket Cloud REST clients behind one GitProvider interface. Plus a 44-entry api-key-provider taxonomy.
hands	Cross-platform computer-use agent. Mouse, keyboard, screen.
install-kit	curl-pipe-bash template for self-hosted Docker apps.
pgflex	One Postgres API. Two modes (real PG ↔ PGlite WASM).

---

Built by Sprayberry Labs

This is one of the open-source building blocks from Sprayberry Labs — an independent studio (Atlanta, GA) that ships bespoke software and fixed-price code & security audits, delivered with the AI workforce these tools are part of.

Part of the askalf ecosystem — a self-hosted AI workforce platform, now in early access.

Got a codebase that needs an expert read? → Scan a repo — free mini-audit, or see the $1,500 fixed-price Audit and build Sprints. · sprayberrylabs.com · hello@sprayberrylabs.com