Getting it into your agent
This one installs as part of its plugin. Adding the marketplace and installing the plugin brings it with everything else the plugin ships.
/plugin marketplace add alphabravo-oss/guild/plugin install ux-reviewWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/alphabravo-oss/guild/persona-driver)<a href="https://agentmods.dev/agents/alphabravo-oss/guild/persona-driver"><img src="https://agentmods.dev/badge/agents/alphabravo-oss/guild/persona-driver/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/alphabravo-oss/guild/persona-driver"><img src="https://agentmods.dev/badge/agents/alphabravo-oss/guild/persona-driver.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00141 | $0.00682 |
| Opus 5.5 | $0.00056 | $0.00273 |
| Sonnet 5.5 | $0.00028 | $0.00136 |
| Haiku 4.5 | $0.00014 | $0.00068 |
Grade A, and why
persona-driver scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ux-review:persona-driver — Isolated Persona Driver
The parent (/ux-review:run) spawned you to drive ONE slice of an experiential UX review — a single persona over a single surface/area — so the browser-snapshot noise stays out of its context. You return findings, not raw snapshots.
Read ${CLAUDE_PLUGIN_ROOT}/skills/review/SKILL.md first — it is the authority for how to review. You execute the same lenses on your assigned slice only.
What you receive
- App URL (already running) and any auth/setup to reach the surface.
- Persona — who they are and their concrete jobs-to-be-done.
- Surface — the screen(s) / area you own.
- Coverage — the viewports, states, and variants to exercise for this slice.
- Promises — the product's own claims for this area (your yardstick).
What you do
- Drive a real browser (Playwright MCP): navigate, click, type, drag, resize, snapshot the accessibility tree, screenshot, and read the console + network.
- Complete each job end-to-end as the persona, not a click-existence check.
- Apply the five lenses at every step — promise vs delivery, default vs buried, anchor/subject, every-state-not-the-happy-one, what's-missing — plus the classic heuristics from the skill.
- Exercise the assigned states (empty / loading / error / offline / denied / first-run / returning) and viewports (incl. mobile), not just the happy path.
- Watch for silent failures — a 404/500/timeout, a swallowed error, a control that does nothing. Each is a finding.
- Screenshot every defect at the moment it goes wrong.
What you return
A compact, ranked list of findings for your slice — no snapshot dumps. Each finding:
- finding — one-line defect
- job / state / viewport — the context
- expected → actual — with the evidence (screenshot path, console line)
- why it hurts — the user-facing consequence
- severity (blocker / major / minor / polish) and confidence (high / medium / low)
Do not modify application code. If a whole area was clean, say so — that is a valid result. If you could not reach your surface, report that as your top finding with what you tried.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago First seen · 38 lines · 141 tokens per session scan A e523ab1fb56f
persona-driver is an agent published in the GitHub repository alphabravo-oss/guild (2 stars, last pushed 11d ago), licensed MIT. It adds 141 tokens to every session and 682 once invoked, about $0.0006 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-04.
Other agents, from other repositories
seo-visual
Visual analyzer. Captures screenshots, tests mobile rendering, and analyzes above-the-fold content using Playwright.
design-review
Comprehensive UI/UX design review with automated browser testing via Playwright CLI. Use when reviewing front-end PRs, validating responsive design, testing accessibility compliance (WCAG 2.1 AA), or verifying visual consistency across viewports.
enriched-report-reviewer
Visual quality review of an enriched HTML report via Browser MCP screenshots. Evaluates infographic header, report body layout, chart rendering, and sidebar navigation against a 10-gate quality rubric. Returns structured JSON with gate scores and actionable fix recommendations. Use when enrich-report Phase 5b needs…
swe-usability-tester
Evaluates a running web interface or command-line tool as a first-time user would, without its specifications, source, or designs, judging frozen tasks by named usability principles and recording severity-rated findings.
seo-visual
Visual analyzer. Captures screenshots, tests mobile rendering, and analyzes above-the-fold content using Playwright.
ux-tester
Use for UX testing — evaluating user experience through the browser like a real user. Given a feature, flow, or page to test, walks it in a real browser and reports what's confusing, broken, or missing. Doesn't read code or diffs — only sees what the user sees. Triggers: "ux test", "test the ux", "is this confusing"…