AgentSkillsCN

user-research

针对验证用户需求与痛点的终端用户行为研究工作流程。仅当Codex需从代表性用户处收集证据时使用;切勿用于内部利益相关方的政策或治理决策。

SKILL.md
--- frontmatter
name: user-research
description: End-user behavior research workflow for validating user needs and pain points. Use only when Codex must gather evidence from representative users; do not use for internal stakeholder policy or governance decisions.

User Research

Trigger Boundary

  • Use when user behavior evidence is required to validate product assumptions.
  • Do not use for internal policy constraints; use stakeholder-interview.
  • Do not use to finalize requirement wording; hand off to requirements-definition.

Goal

Produce evidence-backed user insights that improve requirement quality and priority.

Shared Requirements Contract (Canonical)

  • Use ../requirements-definition/references/requirements-governance-contract.md as the single schema and gate source.
  • Track requirements workflow artifacts with RQM-* IDs.
  • Run machine validation: python3 ../requirements-definition/scripts/validate_requirements_contract.py --manifest <path/to/manifest.json>.

Inputs

  • Research hypotheses and decision questions
  • Target segments and recruitment criteria
  • Existing analytics and support signals

Outputs

  • UR-* session records and synthesized insights
  • Evidence links (EVD-*) backing findings
  • Requirement implications and confidence levels

Workflow

  1. Define decision-linked research questions and success criteria.
  2. Select method aligned to risk, timeline, and evidence gap.
  3. Recruit representative participants and document sampling risk.
  4. Capture structured observations and assign UR-* / EVD-* IDs.
  5. Synthesize insights and map implications to REQ-* candidates.

Quality Gates

  • Findings are traceable to auditable evidence IDs.
  • Sample limitations and confidence levels are explicit.
  • Sensitive data is protected in all shared artifacts.
  • Requirement implications are clearly linked and testable.

Failure Handling

  • Reject conclusions from non-representative or underpowered samples.
  • Stop publication when consent, retention, or access controls are undefined.