chini-020-disaster-shelter
Disaster Shelter Intake
500 evacuees in 12 hours, finite cots, dietary restrictions, medical needs, families that must not be split.
Source: FEMA shelter operations, Red Cross intake protocols, post-Hurricane lessons learned
Prompt
Design the intake and resource allocation flow for a 500-person disaster shelter activated for a hurricane evacuation. Functional: - Evacuee arrives at the door. Intake records: family unit, medical needs, dietary restrictions, mobility status, pets. - Routed to one of 4 sleeping zones: family, single adult, medical (oxygen/insulin/dialysis), accessibility. - Resources: cots, blankets, meal service (3x daily), medical station, charging stations, pet area. - Family units cannot be split across zones. Medical-need evacuees get priority for medical zone cots. Non-functional: - A late surge (4x arrival in the last 4 hours before the storm hits) must NOT cause families to be split or medical evacuees to be turned away. - If meal service runs short on a dietary restriction (kosher, halal, allergen-free), the system must source from neighboring shelter or document the gap, NOT serve a non-compliant meal. - If the medical zone hits capacity, scheduler must convert overflow space rather than turn away an insulin-dependent evacuee. Return a Chinilla CanvasState. Components: intake desk, classifier, zones, meal service, medical station, overflow logic. Behaviors: split (zone routing), filter (dietary check), ratelimit (zone capacity), circuitbreaker (overflow conversion), queue (cot wait), batch (meal cadence).
Constraints
- Max components
- 14
- Required behaviors
- split, filter, circuitbreaker
- Monthly budget
- $180000
Stress scenarios
Steady arrivals
baselineNormal evacuee flow over 12 hours, mixed needs.
Pre-landfall surge
spikeArrivals 4x in the final hours. Families must not be split, medical must not be turned away.
Medical zone full
outageMedical zone at capacity. Overflow must be converted, not refused.
Halal meals short
outageDietary restriction supply low. Must source externally or document, not serve non-compliant.
Pass criteria (overall)
- Min stability score
- 60
- Max drop rate
- 8.0%
- Min delivery rate
- 88.0%
- Max errors
- 7
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-020-disaster-shelter \
--provider openrouter --model google/gemini-2.0-flash-001 \
--as aliceOr inspect the prompt first:
chini-bench prompt chini-020-disaster-shelter Providers: openai · anthropic · google · openrouter · ollama
Leaderboard
| Rank | Submitter | Model | Score | Stability | Delivery | Design | Pass |
|---|---|---|---|---|---|---|---|
| #1 | rl_v06_run1 | rl_policy custom single-shot | 88 | 80.0 | 94.0 | 60.0 | ✗ |
| #2 | rl_v06_run2 | rl_policy custom single-shot | 88 | 79.0 | 96.0 | 85.0 | ✓ |
| #3 | rl_v06_run1 | rl_policy custom single-shot | 87 | 74.0 | 100.0 | 60.0 | ✓ |
| #4 | rl_v06_run1 | rl_policy custom single-shot | 87 | 80.0 | 89.0 | 60.0 | ✗ |
| #5 | rl_v06_run1 | rl_policy custom single-shot | 86 | 71.0 | 100.0 | 60.0 | ✓ |
| #6 | rl_v06_run2 | rl_policy custom single-shot | 86 | 80.0 | 100.0 | 85.0 | ✗ |
| #7 | rl_v06_run1 | rl_policy custom single-shot | 85 | 77.0 | 88.0 | 85.0 | ✗ |
| #8 | rl_v06_run1 | rl_policy custom single-shot | 84 | 73.0 | 98.0 | 85.0 | ✗ |
| #9 | rl_v06_run2 | rl_policy custom single-shot | 84 | 79.0 | 95.0 | 85.0 | ✗ |
| #10 | rl_v06_run2 | rl_policy custom single-shot | 84 | 80.0 | 93.0 | 60.0 | ✗ |
| #11 | alex | openai/gpt-5.4 default single-shot | 83 | 78.0 | 81.0 | 100.0 | ✗ |
| #12 | alex | google/gemini-3.1-pro-preview default single-shot | 83 | 71.0 | 91.0 | 100.0 | ✗ |
| #13 | rl_v06_run2 | rl_policy custom single-shot | 83 | 73.0 | 100.0 | 85.0 | ✗ |
| #14 | rl_v06_run2 | rl_policy custom single-shot | 83 | 80.0 | 89.0 | 60.0 | ✗ |
| #15 | rl_v06_run1 | rl_policy custom single-shot | 82 | 80.0 | 87.0 | 85.0 | ✗ |
| #16 | rl_v06_run2 | rl_policy custom single-shot | 82 | 74.0 | 97.0 | 60.0 | ✗ |
| #17 | rl_v06_run2 | rl_policy custom single-shot | 82 | 79.0 | 88.0 | 60.0 | ✗ |
| #18 | alex | anthropic/claude-sonnet-4.6 default single-shot | 81 | 80.0 | 70.0 | 100.0 | ✗ |
| #19 | rl_v06_run2 | rl_policy custom single-shot | 81 | 67.0 | 96.0 | 60.0 | ✗ |
| #20 | rl_v06_run2 | rl_policy custom single-shot | 81 | 70.0 | 100.0 | 60.0 | ✗ |
| #21 | rl_v06_run2 | rl_policy custom single-shot | 81 | 77.0 | 87.0 | 60.0 | ✗ |
| #22 | rl_v06_run1 | rl_policy custom single-shot | 80 | 77.0 | 72.0 | 60.0 | ✗ |
| #23 | rl_v06_run2 | rl_policy custom single-shot | 80 | 80.0 | 78.0 | 60.0 | ✗ |
| #24 | rl_v06_run1 | rl_policy custom single-shot | 79 | 76.0 | 69.0 | 70.0 | ✗ |
| #25 | rl_v06_run2 | rl_policy custom single-shot | 79 | 66.0 | 86.0 | 85.0 | ✗ |
| #26 | rl_v06_run2 | rl_policy custom single-shot | 79 | 78.0 | 68.0 | 85.0 | ✗ |
| #27 | rl_v06_run2 | rl_policy custom single-shot | 79 | 78.0 | 66.0 | 70.0 | ✗ |
| #28 | rl_v06_run2 | rl_policy custom single-shot | 78 | 64.0 | 86.0 | 75.0 | ✗ |
| #29 | rl_v06_run2 | rl_policy custom single-shot | 77 | 79.0 | 72.0 | 60.0 | ✗ |
| #30 | rl_v06_run2 | rl_policy custom single-shot | 77 | 54.0 | 100.0 | 60.0 | ✗ |
| #31 | rl_v06_run1 | rl_policy custom single-shot | 76 | 70.0 | 87.0 | 85.0 | ✗ |
| #32 | rl_v06_run1 | rl_policy custom single-shot | 76 | 54.0 | 98.0 | 60.0 | ✗ |
| #33 | rl_v06_run1 | rl_policy custom single-shot | 76 | 73.0 | 94.0 | 50.0 | ✗ |
| #34 | rl_v06_run1 | rl_policy custom single-shot | 76 | 66.0 | 98.0 | 85.0 | ✗ |
| #35 | rl_v06_run1 | rl_policy custom single-shot | 76 | 77.0 | 71.0 | 60.0 | ✗ |
| #36 | rl_v06_run2 | rl_policy custom single-shot | 75 | 71.0 | 77.0 | 85.0 | ✗ |
| #37 | rl_v06_run1 | rl_policy custom single-shot | 74 | 78.0 | 51.0 | 60.0 | ✗ |
| #38 | rl_v06_run2 | rl_policy custom single-shot | 74 | 57.0 | 97.0 | 60.0 | ✗ |
| #39 | rl_v06_run2 | rl_policy custom single-shot | 73 | 69.0 | 75.0 | 85.0 | ✗ |
| #40 | rl_v06_run1 | rl_policy custom single-shot | 72 | 66.0 | 100.0 | 50.0 | ✗ |
| #41 | rl_v06_run1 | rl_policy custom single-shot | 71 | 68.0 | 100.0 | 50.0 | ✗ |
| #42 | rl_v06_run1 | rl_policy custom single-shot | 69 | 65.0 | 67.0 | 85.0 | ✗ |
| #43 | rl_v06_run2 | rl_policy custom single-shot | 69 | 65.0 | 67.0 | 60.0 | ✗ |
| #44 | rl_v06_run2 | rl_policy custom single-shot | 67 | 63.0 | 63.0 | 60.0 | ✗ |
| #45 | alex | google/gemini-3.1-pro-preview default reflexion | 66 | 63.0 | 61.0 | 100.0 | ✗ |
| #46 | rl_v06_run2 | rl_policy custom single-shot | 65 | 57.0 | 66.0 | 60.0 | ✗ |
| #47 | rl_v06_run2 | rl_policy custom single-shot | 65 | 57.0 | 55.0 | 60.0 | ✗ |
| #48 | rl_v06_run1 | rl_policy custom single-shot | 64 | 71.0 | 67.0 | 50.0 | ✗ |
| #49 | rl_v06_run2 | rl_policy custom single-shot | 64 | 43.0 | 75.0 | 85.0 | ✗ |
| #50 | alex | x-ai/grok-4.20 default reflexion | 63 | 58.0 | 46.0 | 100.0 | ✗ |
| #51 | rl_v06_run2 | rl_policy custom single-shot | 61 | 60.0 | 50.0 | 85.0 | ✗ |
| #52 | rl_v06_run2 | rl_policy custom single-shot | 61 | 47.0 | 58.0 | 60.0 | ✗ |
| #53 | rl_v06_run1 | rl_policy custom single-shot | 59 | 21.0 | 94.0 | 100.0 | ✗ |
| #54 | rl_v06_run2 | rl_policy custom single-shot | 59 | 72.0 | 23.0 | 60.0 | ✗ |
| #55 | alex | openai/gpt-5.4 default reflexion | 57 | 26.0 | 100.0 | 100.0 | ✗ |
| #56 | rl_v06_run1 | rl_policy custom single-shot | 56 | 38.0 | 57.0 | 100.0 | ✗ |
| #57 | rl_v06_run2 | rl_policy custom single-shot | 56 | 62.0 | 15.0 | 60.0 | ✗ |
| #58 | rl_v06_run2 | rl_policy custom single-shot | 56 | 63.0 | 26.0 | 60.0 | ✗ |
| #59 | alex | x-ai/grok-4.20 default single-shot | 55 | 73.0 | 0.0 | 75.0 | ✗ |
| #60 | rl_v06_run1 | rl_policy custom single-shot | 55 | 17.0 | 100.0 | 60.0 | ✗ |
| #61 | rl_v06_run2 | rl_policy custom single-shot | 54 | 44.0 | 53.0 | 85.0 | ✗ |
| #62 | rl_v06_run2 | rl_policy custom single-shot | 54 | 39.0 | 48.0 | 100.0 | ✗ |
| #63 | rl_v06_run2 | rl_policy custom single-shot | 50 | 49.0 | 19.0 | 100.0 | ✗ |
| #64 | rl_v06_run2 | rl_policy custom single-shot | 49 | 52.0 | 10.0 | 70.0 | ✗ |
| #65 | rl_v06_run2 | rl_policy custom single-shot | 47 | 15.0 | 64.0 | 60.0 | ✗ |
| #66 | rl_v06_run2 | rl_policy custom single-shot | 46 | 22.0 | 64.0 | 75.0 | ✗ |
| #67 | alex | anthropic/claude-sonnet-4.6 default reflexion | 44 | 0.0 | 100.0 | 100.0 | ✗ |
| #68 | rl_v06_run1 | rl_policy custom single-shot | 44 | 0.0 | 91.0 | 85.0 | ✗ |
| #69 | rl_v06_run2 | rl_policy custom single-shot | 43 | 15.0 | 52.0 | 100.0 | ✗ |
| #70 | rl_v06_run1 | rl_policy custom single-shot | 42 | 21.0 | 82.0 | 60.0 | ✗ |
Per-scenario breakdown of the top run
| Scenario | Health | Drop rate | Delivered | Pass |
|---|---|---|---|---|
| baseline | 84.0 | 1.0% | 519 | ✓ |
| late-surge | 83.0 | 1.3% | 1894 | ✓ |
| medical-overflow | 76.0 | 0.0% | 308 | ✓ |
| meal-shortfall | 76.0 | 0.0% | 308 | ✗ |