# Ethical Reasoning

> Ethical Reasoning

- Skill: `sethdford/ethical-reasoning` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add sethdford/ethical-reasoning`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sethdford/ethical-reasoning/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: sethdford (https://skillmd.com/u/sethdford)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/sethdford/ethical-reasoning

---

# Ethical Reasoning

When values collide, use multiple lenses and stakeholder maps instead of a single slogan. Name tensions openly when frameworks disagree—don’t fake certainty.

## When to Use
- Trade-offs affect safety, fairness, privacy, autonomy, or vulnerable groups.
- A choice helps some stakeholders and harms others, or rules conflict with outcomes.

## Behaviors
- **Utilitarian lens**: estimate who gains/loses and how much; watch for discounting minorities or hard-to-measure harms.
- **Deontological lens**: duties, rights, and prohibitions—what must not be done even if aggregate utility rises?
- **Virtue lens**: what habits does this choice reinforce in the team or product culture?
- **Care ethics**: relationships, dependencies, and responsibilities—who is owed care or repair?
- **Justice / fairness**: distribution, procedural fairness, and historical disadvantage; check for proxies that encode bias.
- **Stakeholder map**: list affected parties, mechanisms of effect, and whose voice is missing from the room.
- **Blind-spot scan**: speed, metrics worship, “default allowed,” and power asymmetry between decider and impacted users.
- **Conflict protocol**: if lenses diverge, state the conflict, the non-negotiables, and the least-harm path with transparency.

## Examples
**Example 1:** Feature that boosts engagement but nudges addictive use: map teens vs shareholders; deontological guardrails vs utilitarian metrics; document mitigations.

**Example 2:** Data retention cut: utilitarian efficiency vs user autonomy; care duties to existing customers; publish the fairness trade-off you accepted.