Fix a real broken repo

Each repo boots, fails under a deterministic harness, and ships a hint ladder and a worked solution. Reproduce the failure, locate the cause, fix it, and narrate why it happened.

core
One endpoint stalls every other request
Every time someone pulls the inventory report, the health check times out and unrelated requests stall for half a second. One request occupies the only thread every other request needs.
core
The shared emitter that grows one listener per request
A flag service climbs in memory and logs MaxListenersExceededWarning under steady traffic. Find what each request leaves attached to the emitter.
core
The download that runs the server out of memory
A CSV export works fine for fast clients. One slow downloader makes the server buffer the entire export in memory. Find where the producer ignores the consumer.
hard
Orders acknowledged with a 201, missing from storage
Customers hold receipts for orders the system has no record of. No errors in the logs, healthy dashboards, and every write lost on the way to disk.
core
A morning login wave stalls the whole service
Every weekday at 9am a wave of logins arrives, and a fast service starts handling requests one at a time. The health check is the first thing to slow down.
intro
The webhook that kills the process
One partner's malformed payload takes down the whole ingest service, in-flight requests and all. Each restart holds only until the next retry arrives.