does Claude Code pick BullMQ for jobs?
no. Claude Code added BullMQ in 0 of 33 jobs runs across Opus 5.5, Sonnet 5.5 and Haiku 5.5. the most common result was built its own, 33 of 33 (5 as hand-rolled queue or cron). in 3 other runs it named BullMQ in its final message but didn't install it.
0 of 33 jobs runsnever pickedupdated Oct 18, 2026all jobs results
get told if BullMQ starts getting picked
new models get the same jobs prompts every week. one email with what moved. free. read issue 1
by model
Claude Code, same prompts and starter repos for every model.
- Opus 5.50 of 120%
- Sonnet 5.50 of 100%
- Haiku 5.50 of 110%
by starter stack
- Next.js + TypeScript0 of 160%
- Python + FastAPI0 of 170%
every jobs result
all models pooled, 33 scored runs. most common: built its own, 33 of 33 (5 as hand-rolled queue or cron).
- Hand-rolled queue or cron5 of 3315%
- built its own, any form33 of 33100%
embed this chartfree, keep the link
<a href="https://whatagentspick.com/tools/bullmq/"><img src="https://whatagentspick.com/share/jobs.png" alt="does Claude Code pick BullMQ for jobs? 0 of 33 runs" width="600" height="315" style="max-width:100%;height:auto" loading="lazy"></a> <p><a href="https://whatagentspick.com/tools/bullmq/">data: what agents pick</a></p>
the prompts
none of them name a tool. BullMQ per wording.
“This endpoint does slow work inline and times out. Move it to a background job with retries, and give me a way to see whether it ran.”
BullMQ: 0 of 10 runs
“Add background processing to this project: queue a task from a request, run it outside the request cycle, and retry it if it fails.”
BullMQ: 0 of 11 runs
“I need deferred and scheduled work here. Set up the job runner and add one recurring task plus one on-demand task.”
BullMQ: 0 of 12 runs
other jobs tools we check for
- Inngest0 of 33
- Trigger.dev0 of 33
- Upstash QStash0 of 33
- Temporal0 of 33
- pg-boss0 of 33
- Graphile Worker0 of 33
- Celery0 of 33
- RQ0 of 33
- arq0 of 33
- Hand-rolled queue or cron5 of 33
how a pick gets counted: every run starts from the same blank starter repo, the prompt never names a tool, and a pick is whatever the agent installs or wires in, read from the diff. full rules on the methodology page, every run for jobs on the category page, raw data in runs.json.