AC liquid-neural-network
Build, train, and inspect Liquid Neural Networks (LNNs) — liquid time-constant (LTC) and closed-form continuous-time (CfC) networks with Neural Circuit Policy (NCP) sparse wirings, using the ncps library on PyTorch. Activate when the user asks to build/train a liquid neural network, LNN, LTC, CfC, or NCP model, do continuous-time or ODE-based sequence modeling, or fit small robust recurrent models for time-series prediction. Provides scripts/train_lnn.py (synthetic or CSV time-series training, wiring options, model saving) plus theory and API references. | 构建、训练和检查液体神经网络（LNN）：基于 PyTorch 的 ncps 库，支持液体时间常数（LTC）与闭式连续时间（CfC）网络，以及神经回路策略（NCP）稀疏接线。当用户要求构建/训练液体神经网络、LNN、LTC、CfC 或 NCP 模型，进行连续时间或基于 ODE 的序列建模，或拟合小型稳健的时序预测循环模型时激活。提供 scripts/train_lnn.py（合成或 CSV 时序训练、接线选项、模型保存）以及理论与 API 参考。
Build, train, and inspect Liquid Neural Networks (LNNs) — liquid time-constant (LTC) and closed-form continuous-time (CfC) networks with Neural Circuit Policy…
As a process C 56/100 · Has gaps — weak spots: result and completion, inputs and preconditions, consistency
How to improve
- Your own cases (evals/evals.json, 4–6 real requests with expected answers): the full check would then run those instead of a model-drafted suite.
- A spec.yaml with trigger phrases and assertions — a behaviour contract for CI; `skilltest init` writes a template.
Guard findings · 0
✓ No critical or high findings
Files scanned: 9. Evidence is masked. Grey chips explain why severity was lowered.
Against the Agent Skills spec
- note
frontmatter-keyunknown frontmatter key "homepage"
Process rating: all ten parameters 56/100
- 0Result and completion. Does not say what the result is
- 0Inputs and preconditions. Does not say what the process needs to start
- 0Progress reporting. Says nothing while it works
- 40Consistency. Frontmatter name (liquid-neural-network) differs from the folder (lnn)
- 50Failures and branches. 0 branches, has a failure section
- 60Tools and files. Uses tools (bash, python) that frontmatter does not declare
- 100Steps. 22 steps
- 100When it triggers. States when to use and when not to
- 100Execution cost. Instruction body is 2099 tokens
- 100Running it twice. No mutating operations
Everything here is measured from the skill text rather than judged by a model, so the numbers are checkable. A parameter weighs more when it is a more common reason for the process to stall.
Quality signals
- +5Description has no quoted example phrases that should trigger the skill
- +4Description does not say when NOT to use the skill (false activations)
- +3Output format is not stated: the model decides each time
- +2Single-language instructions
- +3Description length 787: enough signal without eating the budget
- +4Structure: 14 headings
- +3Step-by-step instructions: 22 items
- +4Has examples (5 code blocks)
- +4Reference files are cited in the instructions (2 of 2)
- +3All 2 scripts are documented
- +1License stated
Quality base 70; lint remarks subtract, signals add up to 100. Result: 91.