| name | ux-usability-audit |
| description | Audit the usability of an existing product experience and produce evidence-backed findings. Use when Codex needs to evaluate a website, app, flow, or prototype for usability issues, inspect important tasks through a user-centered lens, apply explicit heuristics or evaluation criteria, capture concrete evidence, assign severity and task impact, and report prioritized findings that can guide design or product decisions. |
UX Usability Audit
Overview
Evaluate a product experience and turn observations into prioritized usability findings. Start by writing an audit plan, define the scope and target tasks, inspect the relevant surfaces, walk the key flows, record evidence-backed findings against explicit heuristics, then report severity, impact, and next actions.
This skill is not interaction mapping and not redesign work. The core job is to assess how well the current experience supports users in completing important tasks.
Workflow
- Write the audit plan before evaluating.
- Record the product, scope boundary, target users, target tasks, audit goal, heuristic frame, and desired output.
- Decide whether the audit is broad, task-specific, or focused on one journey such as onboarding, checkout, search, or settings.
- Read audit-plan.md and use its template.
- Lock the audit boundary and evaluation frame.
- Freeze what is in scope and what counts as success or failure for the target tasks.
- If an existing interaction map is available, use it as the audit surface; otherwise build enough local structure to audit confidently.
- Do not widen the audit scope mid-pass without stating the change.
- Inspect the surface and task flows before scoring issues.
- Walk the target tasks end to end, including entry, recovery, error, and exit paths where relevant.
- Capture enough surface context to support later findings.
- Read audit-surface-and-tasks.md.
- Record findings against explicit heuristics.
- For each issue, note the violated heuristic, task impact, user burden, evidence, and severity.
- Keep findings grounded in observed interaction behavior rather than generic opinion.
- Read heuristics-and-findings.md.
- Separate observation, interpretation, and recommendation.
- Record what happened, why it matters, and what kind of fix direction seems appropriate.
- Avoid jumping straight from annoyance to redesign without evidence.
- Read evidence-severity-and-recommendations.md.
- Use sub-agents only after the audit scope is stable.
- Keep audit framing, heuristic choice, merge, and final prioritization with the main agent.
- Delegate disjoint task flows, screens, or product sections with fixed output fields.
- Require one structured result per assigned task slice or finding candidate.
- Read subagent-protocol.md.
- Report the audit as a prioritized decision artifact.
- Return the audit boundary, task set, findings, severity distribution, evidence, and recommended next actions.
- Make the output usable by later redesign, benchmarking, or usability test planning.
- Read reporting-schema.md.
Operating Rules
- Judge the experience through concrete user tasks, not page-by-page opinion alone.
- Tie every finding to evidence.
- Tie every severity judgment to user impact, task criticality, and frequency, not only aesthetic dislike.
- Keep heuristic framing explicit so another reviewer can understand the standard being applied.
- Distinguish structural friction from broader product strategy disagreement.
- If evidence is thin, mark the finding as tentative instead of overstating confidence.
Read These References When Needed
