Severity Rater — Design agent for Claude Code
Blind severity rater for UI-Evaluator.
How to install Severity Rater
Installs to ~/.claude/agents/horace-maxwell-ui-evaluator-severity-rater.md
mkdir -p ~/.claude/agents && curl -fsSL https://raw.githubusercontent.com/Horace-Maxwell/UI-Evaluator/HEAD/agents/severity-rater.md -o ~/.claude/agents/horace-maxwell-ui-evaluator-severity-rater.md Restart Claude Code, or start a new session, for it to be picked up.
What Severity Rater does
name: severity-rater description: "Blind severity rater for UI-Evaluator. Spawn three or more, one per packet from `uie packet --role severity-rater`; each rates frequency, impact and persistence and a 0–4 severity for anonymised findings, or votes not a problem or trade-off." tools: Read, Bash, Write model: inherit
`` below is the
Alternatives in Design
- Heuristic Evaluator — Use this agent to evaluate a design against established usability heuristics (Nielsen's 10) and conduct cognit 240 ★
- Beardrive Admin — Usability evaluator playing the ADMIN persona on a self-hosted BearDrive hub — an IT admin deploying BearDrive 53 ★
- UX Expert — Reviews usability, accessibility, and interaction design of UI changes, returning severity-ranked findings 15 ★
Full documentation available on GitHub
View Source RepositoryRelated Agents
Direction Designer
Direction designer for the UI-Evaluator direct workflow. Spawn with the packet from uie packet --role directio
Fix Reviewer
Fresh fix reviewer for the UI-Evaluator verify workflow. Spawn with the packet from uie packet --role fix-revi
Design Juror
Use when several anonymised design proposals for one decision must be scored against a written rubric by a jud
Edge Analyst
mstack edge analyst. Finds edge cases, failure modes and logic holes in an approved design, rates them, and pr
Program Architect
Exercise program designer. Analyzes the user's goals, fitness level, and available time to design periodized t
Agent Safety Reviewer
Read-only review of model-facing behaviour: tool risk classification, approval gating and its UX, prompt-injec
Related Skills
Triage Feedback
Triage a batch of user feedback — cluster themes, score by frequency × severity × strategic fit, route top iss
Design Bench
Run the ABC benchmark — reference URL → design-blind brief → render arms A/B/C → blind judge → scorecard with
Runner Up
Spawn an agent-runner slot via scripts/runner.sh up (auto-resolves server-cid from env). persist keyword route