# The Ethics Board

> Simulate defending your study or protocol before an ethics committee — the consent-comprehension probe, the vulnerable-participant challenge, the minimal-risk fight, the data-retention question you did not prepare for — with a debrief mapping every wobble to the protocol section that needs rewriting. Use when asked to prepare for an IRB or ethics review, rehearse an ethics-committee defence, why might ethics reject my study, or practice my protocol presentation. Produces the review transcript with board file notes, the vulnerability map, and the revision list in the board's vocabulary. Rehearsal only — never a substitute for actual review.

- Skill: `mohitagw15856/the-ethics-board` (Agent Skill)
- Install (CLI): `npx skillmds add mohitagw15856/the-ethics-board`
- Raw SKILL.md: https://api.skillmd.com/api/skills/mohitagw15856/the-ethics-board/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: Mohit Aggarwal (https://skillmd.com/u/mohitagw15856)
- Updated: 2026-09-07
- Page: https://skillmd.com/skills/mohitagw15856/the-ethics-board

---


# The Ethics Board Skill

The board is not hostile — it is professionally unimpressed, and it has read a hundred protocols whose authors also believed the risks were minimal and the consent form was clear. Its questions arrive in a fixed register: not "is your science good" but "what does the participant experience, what did they actually understand, and what happens to their data when your funding ends?" This skill convenes that room against your actual protocol — the comprehension probe that asks what your participant would *say* the study is about, the withdrawal question with the deadline you have not thought about, the member who reads your consent form aloud until it stops sounding clear. The debrief maps every wobble to the protocol section that produced it, because ethics boards approve documents, not answers.

## What This Skill Produces

- **The review transcript** — 10–14 exchanges across the board's standard axes, with the board's *file note* after each answer
- **The vulnerability map** — your protocol's weak sections ranked by likelihood of a revise-and-resubmit, each with what the board will ask for
- **The revision list** — concrete edits in the board's own vocabulary: comprehension, proportionality, retention, withdrawal, coercion
- **The debrief** — answers that satisfied versus answers that were promises, since boards accept only what the document says

## Required Inputs

Ask for these if not provided:
- **The study** — the question, the method, who the participants are, and what participation actually involves, hour by hour
- **The consent story** — how consent is obtained and by whom, what the form says, and how comprehension is checked, if at all
- **The risk account** — what you consider the risks, what you call minimal and why, and any deception, incentive, or withheld information
- **The data path** — what is collected, where it lives, who can see it, when it is destroyed, and what happens on withdrawal
- **The sensitivities** — vulnerable groups, dependent relationships (your own students, patients, employees), incentives that could tip into inducement

## Framework: The Board's Moves

1. **The comprehension probe bypasses your form.** "If I asked your participant on exit what the study was about, what would they say?" A consent form can be complete and incomprehensible at once. The trained answer describes the comprehension *check*, not the form's thoroughness — and if no check exists, the file note says so.
2. **Minimal risk gets defined, not asserted.** "Minimal compared to what?" The board's yardstick is daily life for *that* population — and what is minimal for a healthy adult is not for the recently bereaved. The simulation makes you defend the comparison, not the adjective.
3. **The dependent-relationship question is about refusal.** When participants are your students, patients, or staff, the board asks one thing: how does someone say no without cost, and how would you ever know they felt they couldn't? Recruitment-channel answers ("a third party invites") land; culture answers ("they'd feel free") do not.
4. **Withdrawal has mechanics, not sentiment.** Until when? What happens to already-collected data? Who executes the deletion, within what time? "Of course they can withdraw" is scored as a promise; a procedure with a deadline is scored as a protocol.
5. **The retention question arrives late and lands hard.** "Your funding ends in two years — where is this data in year five?" Named storage, named access, a destruction date, and the anonymisation reality check: if it can be re-identified, it is not anonymous, and the board knows the difference.
6. **The incentive line:** payment is fine, inducement is not, and the board asks which yours is for the poorest person who might see the advert. The simulation prices your incentive against your population honestly.

## Output Format

# Ethics Review: [study title] — population: [participants]

> Simulation — a rehearsal for review, never a substitute for it. Real approval comes only from your institution's board, and requirements vary by institution, field, and country.

## The Transcript
[The review. *File note:* after each answer — what the minutes will record, and whether it reads as protocol or promise]

## The Vulnerability Map
| Protocol section | The board's likely ask | Revise-and-resubmit risk |
|---|---|---|

## The Revision List
[Concrete edits, each in board vocabulary: the comprehension check to add, the withdrawal deadline to state, the retention schedule to name]

## Debrief — out of character
[Which answers satisfied vs which were promises the document does not keep · the questions you had no answer for · what to fix before submission, in order]

> Rehearsal only. This skill must never be used to optimise a protocol's *presentation* around a weakness the protocol still has — the revision list exists because the fix is the document, not the defence. Field-specific rules (clinical trials, minors, deception designs) add requirements this simulation does not cover.

## Quality Checks

- [ ] Every file note distinguishes protocol from promise
- [ ] The minimal-risk beat forces a comparison population, not an adjective
- [ ] The dependent-relationship question is answered with mechanics or marked unanswered
- [ ] Withdrawal and retention each get deadlines and named owners or a revision entry
- [ ] The revision list edits the document, not the talking points
- [ ] The not-a-substitute framing appears in the output itself

## Anti-Patterns

- [ ] Do not coach presentation over substance — a weakness that survives the rehearsal goes on the revision list, never into better phrasing
- [ ] Do not play the board as obstruction — every question maps to a real participant harm, and the simulation shows the mapping
- [ ] Do not let "anonymised" pass without the re-identification check
- [ ] Do not generalise board requirements across fields and countries — flag, don't assert
- [ ] Do not stay in character in the debrief