@devleo10/nightwatch-bullmq

Wrap a BullMQ processor so a failed job can be retried, delayed, dead-lettered, or escalated. Rules answer the errors they know. TypeSafe Jev answers the rest. Install path replaces old @devleo10/nightwatch/bullmq snippets; see migration.md.

See what it would do with the failed jobs you already have. Read only:

TYPESAFE_API_KEY=... npx @devleo10/nightwatch-bullmq scan redis://localhost:6379 --queue emails

Then wrap the worker:

npm install @devleo10/nightwatch-bullmq bullmq

import { Worker } from "bullmq"
import { CascadeProvider, JevProvider, RulesProvider, withTriage } from "@devleo10/nightwatch-bullmq"

new Worker("emails", withTriage(async (job) => {
  await sendEmail(job.data)
}, {
  classifier: new CascadeProvider([
    new RulesProvider(),
    new JevProvider({ headers: { authorization: `Bearer ${process.env.TYPESAFE_API_KEY}` } }),
  ]),
  redact: { keys: ["token", "authorization", "email"] },
}))

No key? Pass new RulesProvider() on its own. Dry run defaults to true. The job does not change until policy.dryRun is false.

Guide: https://github.com/devleo10/nightwatch/blob/main/AGENTS.md