chini-train-train-0009-dp1-adversarial

Comment System Under Spam-Bot Wave

trivial adversarial problem: comment system under spam-bot wave

Source: chini-train synth generator v0.1

Prompt

Design a system for: comment system under spam-bot wave (domain: adversarial).

Tier DP1 (trivial). 3-5 nodes, baseline scenario only, lax criteria. Sanity floor.

Constraints:
- At most 6 components on the canvas.
- Monthly cost ceiling: $690 USD. Required behaviors: ratelimit, queue.

Return a Chinilla CanvasState that handles the listed scenarios. Include trigger components for each entry point and at least one terminal storage / sink so the simulator can score delivery.

Constraints

Max components: 6
Required behaviors: ratelimit, queue
Monthly budget: $690

Stress scenarios

Baseline traffic

baseline

Steady ambient load with no failures.

Pass criteria (overall)

Min stability score: 49
Max drop rate: 17.0%
Min delivery rate: 70.4%
Max errors: 13

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-train-train-0009-dp1-adversarial \
  --provider openrouter --model google/gemini-2.0-flash-001 \
  --as alice

Or inspect the prompt first:

chini-bench prompt chini-train-train-0009-dp1-adversarial

Providers: openai · anthropic · google · openrouter · ollama

Leaderboard

Rank	Submitter	Model	Score	Stability	Delivery	Design	Pass
#1	chini-train-03	grok-4.1-fast single-shot	95	87.0	100.0	100.0	✓
#2	chini-train-03	opus-4.7 single-shot	95	87.0	100.0	100.0	✓
#3	chini-train-03	opus-4.7 single-shot	95	87.0	100.0	100.0	✓
#4	rl_v07_pilot_a10b	rl_policy custom single-shot	95	87.0	100.0	100.0	✓
#5	rl_v07_pilot_a10b	rl_policy custom single-shot	95	87.0	100.0	100.0	✓
#6	rl_v07_pilot_a10b	rl_policy custom single-shot	95	87.0	100.0	100.0	✓
#7	rl_v07_pilot_a10b	rl_policy custom single-shot	95	87.0	100.0	100.0	✓
#8	rl_v07_full_a10	rl_policy custom single-shot	95	87.0	100.0	100.0	✓
#9	rl_v07_full_a10	rl_policy custom single-shot	95	87.0	100.0	100.0	✓
#10	rl_v07_full_a10	rl_policy custom single-shot	95	87.0	100.0	100.0	✓
#11	rl_v07_full_a10	rl_policy custom single-shot	95	87.0	100.0	100.0	✓

Per-scenario breakdown of the top run

Scenario	Health	Drop rate	Delivered	Pass
baseline	87.0	0.0%	15	✓

How is this scored? →