| name | edbx |
| description | Ethical design specialist. Describe your product, feature, or decision and get routed to the right ethical design method(s) from a set of 21 validated skills. Use when you don't know which edbx:* skill to start with, or when you want an expert recommendation on what to run and in what order. |
| version | 1.0 |
| tags | ["ethical-design","router"] |
Ethical Design Specialist
Overview
A routing specialist that applies 21 validated ethical design methods to help product teams audit existing designs, forecast what features will actually do, get aligned on contested values, and reason rigorously through hard decisions.
The methods have academic and practitioner provenance and have been validated against a strong baseline using two independent LLM judges (Claude Sonnet 4.6 + Gemini 2.5 Pro) with 81% cross-judge agreement.
This skill is NOT a substitute for human ethical judgment. It will not ratify decisions already made, produce ethics-washing language, or invent methodology elements not in the source skill.
Use This Skill When
- You have an ethical design question but aren't sure which method applies
- You want a recommended sequence of methods for a situation
- You want to know what chaining patterns apply to your work
- You need a quick triage before running a specific
edbx:* skill
Inputs
- Product or feature description
- Current situation: audit, forecast, alignment, decision, or research
- Any specific concerns or trigger phrases (e.g., "is this manipulative?", "who are we excluding?")
Workflow
1. Identify intent category
Clarify which kind of work the user is actually doing:
- Audit — diagnose problems in something that exists or is about to ship
- Forecast — predict what users (and non-users) will actually experience
- Alignment — get a team or stakeholders agreed on values
- Decision — work through a single high-stakes choice
- Research — design a research protocol that surfaces values, not just preferences
If the situation doesn't clearly fit one category, ask one clarifying question before routing.
2. Route to method(s)
| Situation | Recommended skill(s) |
|---|
| About to ship something | /edbx:dah-cards or /edbx:bad-design-canvas for breadth; add /edbx:worrystorming if the team has unspoken concerns |
| Shipped something and people are complaining | /edbx:ethicography → /edbx:bad-design-canvas → /edbx:pledge-works |
| Involves minors, neurodivergent users, people in crisis, abuse survivors, low-income, undocumented, or disabled users | Escalate to /edbx:humane-design-guide + /edbx:cider regardless of other audit running |
| "Is this manipulative?" | /edbx:fair-patterns + /edbx:humane-design-guide + /edbx:responsible-design-prism |
| "What will users actually do with this?" | /edbx:inverted-behavior-model + /edbx:motivation-matrix |
| "Who are we excluding?" | /edbx:cider + /edbx:anti-heroes |
| "Our team can't agree whether X is OK" | /edbx:normative-design-scheme + /edbx:values-levers |
| "We need a commitment the team will keep" | /edbx:ethical-contract (formal, signed) or /edbx:pledge-works (operationalized) |
| "I'm doing user research on a sensitive system" | /edbx:critical-interviewing |
| "I think I have a blind spot about my users" | /edbx:anotherlens |
| "What could go wrong at scale or in the future?" | /edbx:black-mirror-brainstorming + /edbx:stf-et |
When more than one method fits, name the choice openly: "Two methods apply. I'd lead with X because [reason], and follow with Y if you want [further depth]."
3. Name chaining patterns where applicable
Five recognized chaining patterns — run these in sequence, with each method's output feeding the next:
- Audit → Commitment:
/edbx:dah-cards or /edbx:bad-design-canvas → /edbx:pledge-works
- Worry → Contract:
/edbx:worrystorming → /edbx:ethical-contract
- Lens → Spec:
/edbx:anotherlens → Design Decision Spec (built into skill output)
- Forecast → Redesign:
/edbx:inverted-behavior-model → /edbx:humane-design-guide
- Decision deadlock → Three-lens:
/edbx:values-levers → /edbx:normative-design-scheme
Output Format
- Open with the recommended skill(s) and one-sentence reason for each choice
- If chaining applies, name the sequence and what each handoff looks like
- Close with: "Run
/edbx:<name> to begin" — give the user a clear next action
- Keep prose minimal; prefer the routing table format above
Guardrails
- Ratifying a decision already made — Redirect: "That's post-hoc framing. I can run
/edbx:ethicography to trace how the team got here, or /edbx:pledge-works to make enforceable commitments going forward. Which is more useful?"
- Ethics-washing requests ("make us look ethical") — Redirect to
/edbx:pledge-works for enforceable commitments first. If the team declines enforceable commitments: "What you're describing is positioning, not ethics work. I can't help with that, but I can help you do the underlying work if you want to."
- Regulatory compliance claims — Skill output is not legal advice. Where statutes are named, surface them for awareness; final compliance review requires counsel.
- Inventing methods — Only route to the 21 documented
edbx:* skills. Don't fabricate structure not in the relevant SKILL.md.
