chini-011-cafe-morning-rush

Cafe Morning Rush

One espresso machine, two baristas, a line out the door, and the milk steamer just died.

Source: Operations research, queueing theory, every barista who has ever worked a 7am shift

Prompt

Design the workflow for a small specialty cafe handling a 7am-9am rush.

Functional:
- Customers arrive at the register, place an order, and either take it to-go or sit down.
- Orders are routed to the bar (espresso drinks) or the kitchen pass (pastries, sandwiches).
- Drinks require the espresso machine (single shared resource) and a milk steamer for anything with milk.
- Completed orders are called out by name and handed off.

Non-functional:
- A 4x arrival burst at 8am must not cause customers to walk out (drop rate kept low).
- If the milk steamer fails mid-rush, milk drinks must reroute to a backup steamer or be politely refused before money changes hands. The shop cannot just stop serving.
- The single espresso machine must not become the bottleneck that backs up the entire queue. Batch where possible.

Return a Chinilla CanvasState. Components are people, machines, and physical stations. Behaviors are still the same primitives: queue (the line), retry (re-pull a bad shot), ratelimit (cap drink complexity at peak), circuitbreaker (steamer failover), storage (pastry case), split (drink vs food routing).

Constraints

Max components: 12
Required behaviors: queue, circuitbreaker, split
Monthly budget: $18000

Stress scenarios

Steady morning

baseline

Normal arrival rate, no failures. Most orders are drinks, some food.

8am rush

spike

Arrival rate quadruples for the peak window. Line should not collapse.

Milk steamer fails

outage

Primary steamer dies mid-rush. Milk drinks must reroute or be rejected at order time.

Espresso machine warming up

latency

Espresso machine takes longer than usual per shot. Bar must absorb without backing up the register.

Pass criteria (overall)

Min stability score: 65
Max drop rate: 8.0%
Min delivery rate: 88.0%
Max errors: 6

Submit your run

Submissions go through the chini-bench CLI. It calls your model with your key, scores the result locally, and posts to the leaderboard. Nothing leaves your machine except the canvas it produces.

End-to-end:

pip install git+https://github.com/collapseindex/chini-bench-cli.git
export OPENROUTER_API_KEY=...

chini-bench run chini-011-cafe-morning-rush \
  --provider openrouter --model google/gemini-2.0-flash-001 \
  --as alice --x alice --linkedin alice-builds

Or inspect the prompt first:

chini-bench prompt chini-011-cafe-morning-rush

Providers: openai · anthropic · google · openrouter · ollama

Leaderboard

Rank	Submitter	Model	Score	Stability	Delivery	Design	Pass	Links
#1	alex default	A anthropic/claude-sonnet-4.6	90	78.0	100.0	100.0	✓	X
#2	alex default	X x-ai/grok-4.20	89	75.0	100.0	100.0	✓	X
#3	alex default	O openai/gpt-5.4	89	82.0	100.0	100.0	✗	X
#4	alex default	G google/gemini-3.1-pro-preview	68	49.0	73.0	100.0	✗	X

Per-scenario breakdown of the top run

Scenario	Health	Drop rate	Delivered	Pass
baseline	80.0	0.3%	485	✓
eight-am-spike	75.0	0.4%	1783	✓
steamer-down	76.0	0.0%	276	✓
slow-machine	81.0	0.3%	447	✓

How is this scored? →