AgentSkillsCN

ethical-technology-use

撰写并组织具有针对性、公平性与可操作性的反馈意见。当用户请求反馈、开展绩效面谈、进行同行评审，或希望在不损害信任的前提下调整行为时，此技能将助您精准把握要点，给出建设性建议。

SKILL.md
--- frontmatter
name: ethical-technology-use
description: Applies ethical and responsible technology practices (privacy, security, consent, transparency, bias, IP) in day-to-day work with AI/tools. Use when using AI on user/client/student data, deploying features, collecting analytics, or generating content.

Ethical Technology Use

Goal

Make responsible choices that protect people, reduce risk, and preserve trust.

Trigger signals

  • •“Can I put this data in an AI tool?” “Is this ok to scrape/log/store?”
  • •Shipping features that affect users (ranking, recommendations, surveillance-like analytics).
  • •Generating content/code with AI where IP, privacy, or attribution matters.

Core workflow (fast checklist)

  1. •Data classification: public / internal / personal / sensitive.
  2. •Consent & purpose: Do we have consent? Is the use aligned with the purpose?
  3. •Minimization: Can we remove or redact identifiers?
  4. •Security: Where is it stored? Who can access? Retention period?
  5. •Fairness: Who could be harmed? Any bias or exclusion risks?
  6. •Transparency: What should we disclose (to users, students, stakeholders)?
  7. •Accountability: Who owns the decision and review?

Practical defaults

  • •Never share sensitive personal data with third-party tools without explicit permission and safeguards.
  • •Prefer “least data, least time”: minimize fields and keep retention short.
  • •Document decisions: one paragraph decision record with risks + mitigations.

Templates

Decision record (one paragraph)

“We will use [tool] for [purpose] with [data type]. Risks: [privacy/bias/security]. Mitigations: [redaction, access control, retention]. Owner: [name]. Review date: [date].”

Transparency line (user-facing)

“We use automated tools to support [task]. You can opt out / request review via [path].”

Output defaults

  • •Provide a go/no-go recommendation with mitigations.
  • •Provide a short policy-style snippet the user can paste into docs.

Guardrails

If legal/compliance is uncertain, recommend consultation with DPO/legal/security and keep advice conservative.