Loading
Governs models and AI systems in production — intended use, evaluation, monitoring, human oversight, documentation, and the decision to deploy or retire. Use this before deploying a model or AI featur
The skill file stays in the publisher's repository.
Paste the copied command in a terminal. It installs this SKILL.md into that agent.
npx skills add cbrock84/headcount --skill ai-ml-governance -a cursor -y
Governs models and AI systems in production — intended use, evaluation, monitoring, human oversight, documentation, and the decision to deploy or retire. Use this before deploying a model or AI feature, when defining evaluation criteria, when a model's behavior has drifted, when assessing AI risk or regulatory exposure, or when deciding whether an AI system is fit for a consequential decision.
Governs models and AI systems in production — intended use, evaluation, monitoring, human oversight, documentation, and the decision to deploy or retire. Use this before deploying a model or AI feature, when defining evaluation criteria, when a model's behavior has drifted, when assessing AI risk or regulatory exposure, or when deciding whether an AI system is fit for a consequential decision.
Regimes governing automated decision-making differ by jurisdiction and sector and are changing quickly. Anything affecting credit, employment, housing, insurance, healthcare, or education carries specific legal obligations — involve Legal & Risk and qualified counsel rather than treating it as an engineering question.
Write down what the system is for, what it is not for, who is affected by its output, and what happens when it is wrong. Most AI failures are use outside intended scope by someone who did not know the scope existed.
Then decide the consequence tier, because it sets everything after it:
A held-out evaluation set that reflects real inputs, including the awkward ones. Built before deployment and kept stable, or you cannot compare versions.
Models degrade silently: the world moves, inputs drift, and accuracy falls without any error being raised.
Monitor input distribution against training, output distribution over time, performance against whatever ground truth arrives later, and the rate of human override. A rising override rate is the best early warning you have, and it is usually already visible in a queue nobody reads.
Meaningful, not nominal. A reviewer approving hundreds of decisions an hour is not overseeing anything — they are laundering the model's output through a person.
Meaningful oversight requires the reviewer to see why the model decided, to have time to disagree, and to have their disagreement change the outcome and be recorded.
Per model: intended use and exclusions, training data and its provenance, evaluation results by segment, known limitations, monitoring in place, and the owner. This is what you need when someone asks why a decision was made — and increasingly what a regulator expects to see.
Have a way to turn it off. Know what happens to the process when you do, and confirm the fallback still works — a manual path that has not been exercised in two years is not a fallback.
references/sources.md in this skill lists the outside authorities that settle the questions
here — what each one is authoritative for, and what you may do with it. Check them before
answering on anything they cover, and cite what you used. Most are free to read and not free
to reproduce; the use note on each is binding.