ux-heuristic-eval

What this skill does

<!-- VERSION: 1.0.0 | DATE: 2026-03-04 | SOURCE: skills/user-experience/SKILL.md | PARENT: /user-experience skill -->

# Heuristic Evaluation Sub-Skill

> **Version:** 1.0.0
> **Framework:** Jerry User-Experience -- Heuristic Evaluation
> **Constitutional Compliance:** Jerry Constitution v1.0
> **Parent Skill:** `/user-experience` (`skills/user-experience/SKILL.md`)
> **Wave:** 1 (Zero-Dependency)
> **Project:** PROJ-022 User Experience Skill | GitHub Issue [#138](https://github.com/geekatron/jerry/issues/138)

## Document Sections

| Section | Purpose |
|---------|---------|
| [Document Audience](#document-audience-triple-lens) | Triple-Lens audience guide |
| [Purpose](#purpose) | Sub-skill overview and key capabilities |
| [When to Use This Sub-Skill](#when-to-use-this-sub-skill) | Activation triggers and scope boundaries |
| [Available Agents](#available-agents) | Single agent with role, model, and output location |
| [P-003 Compliance](#p-003-compliance) | Worker agent hierarchy position |
| [Invoking the Agent](#invoking-the-agent) | Invocation via ux-orchestrator |
| [Methodology](#methodology) | Nielsen's 10 heuristics, severity scale, evaluation workflow, single-evaluator reliability |
| [MCP Dependencies](#mcp-dependencies) | Figma REQ with screenshot-input fallback; Context7 for UX docs |
| [Output Specification](#output-specification) | Output location, L0/L1/L2 structure, required sections |
| [Routing](#routing) | Keywords and lifecycle-stage routing integration |
| [Cross-Framework Integration](#cross-framework-integration) | Handoff to Behavior Design, HEART Metrics, and CRISIS mode |
| [Synthesis Hypothesis Confidence](#synthesis-hypothesis-confidence) | Confidence classifications for heuristic evaluation outputs |
| [Constitutional Compliance](#constitutional-compliance) | Governing principles |
| [Registration](#registration) | H-26(c) parent-routed registration model and AGENTS.md confirmation |
| [Deployment Status](#deployment-status) | Wave 1 stub agent status and implementation timeline |
| [Quick Reference](#quick-reference) | Common workflows and agent selection hints |
| [References](#references) | Full repo-relative paths, requirements traceability, external citations |

## Document Audience (Triple-Lens)

This SKILL.md serves multiple audiences:

| Level | Audience | Sections to Focus On |
|-------|----------|---------------------|
| **L0 (Stakeholder)** | Product managers, designers | [Purpose](#purpose), [When to Use This Sub-Skill](#when-to-use-this-sub-skill), [Quick Reference](#quick-reference) |
| **L1 (Developer)** | Engineers invoking the agent | [Invoking the Agent](#invoking-the-agent), [Methodology](#methodology), [Output Specification](#output-specification) |
| **L2 (Architect)** | Workflow designers, skill maintainers | [Cross-Framework Integration](#cross-framework-integration), [MCP Dependencies](#mcp-dependencies), [Synthesis Hypothesis Confidence](#synthesis-hypothesis-confidence) |

---

## Purpose

The Heuristic Evaluation sub-skill provides structured, evidence-based usability evaluation using Nielsen's 10 usability heuristics. It targets tiny teams (1-5 people) who lack dedicated UX evaluators by providing AI-augmented systematic interface inspection with severity-rated findings and actionable remediation recommendations.

This sub-skill is part of Wave 1 (Zero-Dependency), meaning it requires no prior wave completions and operates with minimal infrastructure. It is the foundation evaluation framework that other sub-skills build upon.

### Key Capabilities

- **Systematic Heuristic Inspection** -- Evaluates interfaces against all 10 of Nielsen's usability heuristics sequentially, ensuring complete coverage
- **Severity-Rated Findings** -- Each finding receives a 0-4 severity rating per Nielsen's severity taxonomy (Not a problem to Usability catastrophe) with evidence from the interface
- **Remediation Recommendations** -- Actionable fix suggestions with effort estimates (Low/Medium/High) for each finding
- **Multi-Screen Evaluation** -- Coordinates evaluation across multiple screens or flows with finding deduplication
- **Model Escalation** -- Default Haiku for high-volume checklist evaluation; auto-escalates to Sonnet for complex evaluations
- **Cross-Framework Feed** -- Findings feed directly into Behavior Design diagnosis and HEART Metrics measurement

> **Source:** Key capabilities derived from parent SKILL.md [Key Capabilities] (line 103: "Heuristic Evaluation -- Nielsen's 10 Heuristics with severity-rated findings (Wave 1)") and [Available Agents] (lines 152, 165).

---

## When to Use This Sub-Skill

Activate when:

- Evaluating an existing interface against established usability principles
- Conducting a structured usability audit with severity-rated findings
- Reviewing interface design quality during the "During design" lifecycle stage
- Performing interface review as part of a comprehensive UX audit (multi-sub-skill route with HEART Metrics)
- Executing CRISIS mode step 1 (identify UX issues via Nielsen's 10 heuristics)
- Evaluating AI product interfaces as an interim path when the AI-First Design sub-skill is not yet deployed
- Needing a baseline usability assessment before deeper behavioral or metrics analysis

Do NOT use for:

- Diagnosing why users fail to take a specific action -- use `/ux-behavior-design` (Fogg B=MAP) instead. Heuristic evaluation identifies design-level issues but does not trace behavioral root causes.
- Measuring quantitative UX health metrics -- use `/ux-heart-metrics` (Google GSM) instead. Heuristic evaluation produces qualitative findings, not metric baselines.
- Accessibility-specific compliance auditing -- use `/ux-inclusive-design` (WCAG 2.2) instead. Heuristic evaluation covers general usability; dedicated accessibility evaluation requires WCAG success criteria.
- Testing hypotheses about design changes -- use `/ux-lean-ux` (Lean UX) instead. Heuristic evaluation assesses existing interfaces, not hypothesis-driven experiments.
- Security-focused interface review -- use `/eng-team` instead. Heuristic evaluation applies usability principles, not security threat models.
- General research without UX evaluation focus -- use `/problem-solving` instead.

> **Source:** Routing logic derived from parent SKILL.md [Lifecycle-Stage Routing] (lines 295-334) and `skills/user-experience/rules/ux-routing-rules.md` [Stage Routing Table], [Common Intent Resolution].

---

## Available Agents

| Agent | Role | Tier | Mode | Model | Output Location |
|-------|------|------|------|-------|-----------------|
| `ux-heuristic-evaluator`** | Nielsen heuristic evaluation specialist | T4 | Systematic | Haiku* | `projects/${JERRY_PROJECT}/engagements/{engagement-id}/ux-heuristic-evaluator-{topic-slug}.md` |

*Haiku for high-volume checklist evaluation; escalates to Sonnet when: (1) critical finding count >= 3 (severity 3 or 4), (2) Figma MCP benchmark fails pre-launch threshold, or (3) evaluation spans > 50 screens. Escalation is automatic within the orchestrator's routing logic per AD-M-009 model selection justification.

**STUB: The agent definition file (`skills/ux-heuristic-eval/agents/ux-heuristic-evaluator.md`) currently contains frontmatter, identity, purpose, and guardrails sections only. Full agent body implementation (`<input>`, `<capabilities>`, `<methodology>`, `<output>` sections) is pending Wave 1 completion of PROJ-022 EPIC-002. The SKILL.md specifies the methodology and output contract that the agent will implement.

**Tool tier:** T4 (External) = Read, Write, Edit, Glob, Grep + WebSearch, WebFetch + Context7 MCP. The T4 tier enables access to external UX standards documentation via Context7 and web search. Bash is intentionally excluded; T4 tier does not require shell access for MCP operations. See `agent-development-standards.md` [Tool Security Tiers] for full tier definitions.

The agent produces output at three levels per AD-M-004:
- **L0 (Executive Summary):**