Plan and run user interviews without losing the evidence trail. Control recruitment, consent, facilitation, session notes, study changes, and the handoff into synthesis.
An interview guide can look finished while the study is still unsafe or impossible to interpret. The wrong people may have been recruited. Consent may not cover recording. A leading question may turn the team's preferred answer into a research finding.
This skill controls the work before, during, and after each session. It defines the product decision, sampled population, recruitment rules, incentive, consent, privacy boundary, guide-to-learning-question map, pilot, protocol versions, and stop conditions. During fieldwork, it keeps participant evidence separate from moderator interpretation and records deviations, missing sections, technical failures, withdrawals, and follow-ups.
The final handoff includes the study contract, recruitment and disposition summary, screener, discussion guide, consent record, participant-coded evidence inventory, coverage gaps, protocol history, and a synthesis brief. It does not claim that interviews prove prevalence, causality, or statistical representativeness.
Input
Plan six interviews with finance admins who corrected at least one failed payout in the last 60 days. We need to understand their current recovery workflow before choosing a redesign. Recruiting comes from customer success, sessions are remote, and recording is optional.
Output
A decision-linked study contract, observable screener, quota and disposition plan, consent and optional-recording script, neutral guide mapped to learning questions, pilot gate, participant-code scheme, evidence-note template, privacy boundary, and ready-to-pilot verdict.
Input
Our sales team sent us ten friendly customers and wants us to record 30-minute calls tomorrow. The guide asks whether they love the new AI assistant. We have no retention rule, incentive policy, or approved prototype-sharing note.
Provide the product decision, target population, research objective, recruitment source, time and incentive constraints, consent and recording rules, data-retention boundary, study materials, and the people responsible for fieldwork and review. Ask the agent to plan the study, review a screener or guide, prepare session materials, or turn completed participant-coded notes into a synthesis-ready fieldwork handoff.
The decision this study may inform, the decision owner, deadline, primary learning objective, and explicit non-goals.
Observable eligibility, exclusions, quotas, recruitment source, incentive, scheduling owner, language or accessibility needs, and any conflict rules.
Approved consent and recording choices, withdrawal route, data minimization, storage, access, retention, deletion, moderator, note taker, and reviewer.
SKILL.md; study contract, recruitment, screener, consent, and risk controls; session, pilot, evidence-note, coverage, and synthesis-handoff templates; OpenAI interface metadata
No reviews yet.
Output
A blocked fieldwork verdict, exact approval and evidence requests, recruitment-bias risks, a rewritten neutral guide, consent and disclosure gaps, a safe pilot path, owners, and the work that can proceed before participants are contacted.
Input
Seven sessions are complete. Two used guide v2, one participant declined recording, one session lost audio after 18 minutes, and a participant asked us to delete their data. Prepare the fieldwork handoff without synthesizing themes.
Output
A participant-coded inventory, consent and deletion states, protocol-version map, partial-session and technical-failure records, evidence coverage by learning question, quota and perspective gaps, allowed evidence set, prohibited claims, artifact index, and an evidence-matched synthesis-readiness verdict.
Creator
Aavrilo
