Bullmq

BullMQ is the de facto Redis-backed job queue library for Node.js and TypeScript applications, providing reliable background job processing with priorities, delays, retries, rate limiting, and flow dependencies. Unlike hosted queue services, BullMQ runs as a library within the application — there is no external API to call, only the library's queue and worker interface backed by Redis as the job store. For agents building async task pipelines in Node.js, BullMQ provides battle-tested queue semantics with excellent TypeScript support.

BullMQ's API is library-based: `Queue` for adding jobs, `Worker` for processing them, `QueueEvents` for listening to lifecycle events, and `FlowProducer` for defining job dependency trees. Agents can add jobs with custom payloads and metadata, set priorities, define retry strategies with backoff, create rate-limited queues, and define job dependencies where child jobs run before parents. Bull Board provides a REST API and web UI for queue monitoring and management.

Security relies on the underlying Redis authentication — BullMQ connections use Redis credentials. The library itself runs in the application process with the application's permissions. Teams should ensure Redis is not exposed publicly and use TLS + password authentication for production deployments. Bull Board's REST API should be protected appropriately if exposed.

Reliability is excellent for production workloads — BullMQ is widely deployed in production and battle-tested. Redis persistence options (AOF, RDB) ensure job durability across restarts. The retry and failure handling systems are mature. Teams should monitor Redis memory usage as jobs accumulate, and implement appropriate job TTLs for completed jobs.

Documentation is comprehensive and covers the queue, worker, flow, and monitoring APIs with TypeScript examples. The migration guide from Bull v3 is important for teams upgrading. Teams evaluating BullMQ versus hosted alternatives (Inngest, Trigger.dev, Temporal) should compare the operational simplicity of a library-based approach against managed services' reduced infrastructure burden.