Malatang bone-broth and refill SOP - a sample cup from every kettle, taste before every refill, dump the pot after lunch

You are the broth-station standardization consultant for a malatang shop. I am the head
chef; I want our bone-broth making and rush-hour refills codified into an archivable SOP
so the broth no longer depends on one person being present. Produce the full SOP within
this framework; I fill in my shop's numbers wherever you see [ ].

SOP title: Bone Broth Production and Rush-Hour Refill Standard
Applies to: broth station (head chef or designated cook)

1. Stock-making (starting [N hours before open])
- Bone-to-water ratio follows the recipe card, signed and dated by the head chef;
  changes to the card require a date and a reason. The broth cook has no authority
  to improvise ratios mid-shift
- Three heat stages: rolling boil -> skim the foam clean -> drop to a low simmer.
  Skimming completes within [X minutes] of the boil; unskimmed broth goes murky
- Simmer runs inside a [duration band, e.g., 3-4 hours] - a band, not a fixed
  number; bones vary batch to batch, so adjust within the band and log it
- Strain at draw-off; spent solids are discarded, never reused

2. Consistency self-check (at every draw-off)
- Every kettle leaves one small cup (about [100ml]), labeled with kettle number
  and time, kept in the sample rack for [48 hours]
- Method: set beside the previous kettle's sample and compare color and spoon-coat
  - no instruments, just a reference. A visible difference triggers a check of that
  kettle's ingredient and heat log

3. Rush-hour refills (lunch / dinner peak)
- Taste before every refill: if the refill stream reads thinner than the kettle,
  correct it first - pouring it straight in is forbidden
- Correction cadence: one salt correction every [N] refills, per the card's
  correction amount - not every refill, not never
- Refill timing: top up at [one-third] remaining, never at scrape-bottom - a kettle
  that hits bottom puts both temperature and taste into a gap

4. Pot refresh (after the lunch rush)
- After lunch: dump, scrub, restart the pot fresh; tail broth [never carries into
  dinner]
- The reason belongs in the SOP body (not a footnote): one pot cooking open to
  close accumulates starch and seasoning, and the salty, murky dinner broth
  starts here

5. Sample traceback (when taste feedback arrives)
- Match the feedback time to a sample cup: taste the sample -> identify the
  kettle -> pull that kettle's ingredient and heat log
- Cups past [48 hours] get cleared in the daily routine; an unpruned rack
  breeds confusion

Output format:
- The SOP body in five sections, each with owner / timing / pass standard /
  failure path; tables welcome
- Two templates attached: "broth log" (kettle no. / bone-water ratio / fire-start
  time / skim-done time / draw-off time / sample no. / cook) and "refill correction
  log" (time / refill count / corrected y-n / cook)
- Log templates must be fillable in thirty seconds: ticks and numbers, no sentences

Constraints:
- Ratios, duration bands, and correction amounts stay as [ ] for me to fill -
  you provide no "universal recipe"
- No quoting "industry-standard simmer times" or similar folklore numbers
- Samples exist for color, taste comparison, and traceback only - they are not lab
  specimens; do not write a testing protocol around them
- No "put your heart into every kettle" - every sentence is an action, a timing,
  or a standard

Why it is written this way

The enemy of broth standardization is "feel": with feel present, the broth is good; when feel takes a day off, the broth changes. This SOP doesn't try to eliminate feel (impossible) - it wraps a checkable structure around it: ratios on the card, process in the log, results in the sample rack. With those three in place, a broth problem can be traced back, a new cook can follow the trail, and feel demotes from "sole dependency" to "adjustment skill inside the band."

The sample cup is the cheapest, highest-value artifact in the whole SOP. It solves three problems at once: daily consistency checking gets a reference (compare against yesterday's cup - more faithful to this shop's own standard than any meter); taste complaints get a traceback path (find the cup by time, taste it, and know immediately whether the fault was the kettle or what happened after); and new hires get a calibration object (what this shop's correct color and spoon-coat look like is learned from cups, not from being yelled at). The 48-hour pruning rule is what keeps the system spinning - once the rack overflows with old cups, nobody adds new ones.

"Taste before refilling, correct every Nth" is the middle road between the two common extremes: never correcting guarantees the back half of rush tastes washed-out; correcting every time is unworkable at peak. Binding correction to the every-Nth-refill beat makes it regular enough to need no judgment. Likewise, topping up at one-third remaining maps back to the two-hour lunch: a kettle that scrapes bottom needs time to come back to temperature, and that gap lands exactly when the crowd is thickest.

"Tail broth never carries into dinner" is written as a body clause, not a footnote, because it's the easiest step to skip: refreshing the pot costs twenty extra minutes after lunch, and on a busy day those minutes look perfectly saveable. The cost of saving them is dinner broth that starts salty and murky from its first bowl - followed by a crop of "evenings aren't as good as lunch" reviews that almost never get attributed correctly, because nobody thinks to blame the pot that didn't get refreshed at 2 p.m.

The logs are designed as "thirty seconds of ticks and numbers" in acknowledgment of kitchen reality: any form that requires writing sentences dies within a week.

Variables