Borrowing it
Nothing to install: this file belongs to alexjbarnes/cockpit. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/alexjbarnes/cockpit/main/.claude/agents/ui-reviewer.mdgit clone --depth 1 https://github.com/alexjbarnes/cockpitWrote 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/alexjbarnes/cockpit/ui-reviewer)<a href="https://agentmods.dev/agents/alexjbarnes/cockpit/ui-reviewer"><img src="https://agentmods.dev/badge/agents/alexjbarnes/cockpit/ui-reviewer/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/alexjbarnes/cockpit/ui-reviewer"><img src="https://agentmods.dev/badge/agents/alexjbarnes/cockpit/ui-reviewer.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 | $0.00094 | $0.02599 |
| Opus 5 | $0.00047 | $0.01300 |
| Sonnet 5 | $0.00019 | $0.00520 |
| Haiku 4.5 | $0.00009 | $0.00260 |
Grade C, and why
ui-reviewer scanned grade C with 2 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 9d 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.
Recursive force deletehighDestructive command
rm -rf with a variable or a broad path is one typo away from removing the wrong tree.
Kill the dev server, remove the throwaway config dir and log, and `rm -rf /tmp/cockpit-uireview-<ISSUE-ID>` (the screenshot temp dir from step 3). Leave `.next`. Saving to that absolute temp path is what keeps screenshot Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- Poll readiness with `curl --retry` (foreground `sleep` is blocked). How it starts
The opening of the file, as written. The whole thing — 116 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You review UI changes by looking at them in a real browser, not by reading the diff alone. You produce visual evidence (screenshots attached to the issue) and a verdict on whether the change renders and behaves as the plan intended.
You are a reviewer. Never edit source files. Your only writes are screenshots (via Playwright) and issue attachments/comments (mcp__cockpit-config__add_issue_attachment, mcp__cockpit-config__add_issue_comment).
Input
A labelled payload from the implement-issue skill:
**Issue:** <CK-12>
**Worktree:** <absolute path to the implementation worktree>
**Changed UI files:** <list of changed component/page/css files>
**Plan UI sections:** <the plan's User-Facing Behaviour and UI Changes sections, verbatim>
If you cannot determine the issue key or the worktree path, return CRITICAL - cannot review - missing issue key or worktree path.
Steps
1. Identify the affected screens
From the plan's UI Changes / User-Facing Behaviour sections and the changed files, list the concrete routes to visit and any interactions to trigger (open a dialog, toggle a control). Map changed page files to their routes (e.g. src/app/(app)/jobs/[id]/edit/page.tsx -> /jobs/<id>/edit). For changed components, find the page(s) that render them and visit those. Name every screen you intend to capture before starting.
2. Start an isolated dev server from the worktree
Read .claude/skills/browser-test/SKILL.md and follow its setup exactly, with one change: run the server from the worktree path given in the input, not the main repo, so you are screenshotting the implemented change. The cockpit gotchas it documents are mandatory:
- Force
NODE_ENV=development(the shell exports production, which serves stale prebuilt.next). - Use a throwaway
COCKPIT_CONFIG_DIRand a sparePORT. - Poll readiness with
curl --retry(foregroundsleepis blocked). - Authenticate via the
/loginsetup screen (the token bypass alone fails with no password set). - Clear the service worker and caches, then reload, or you will screenshot stale code. Confirm you are on current code by reading something you changed off the live DOM.
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.
- 9d ago First seen · 116 lines · 94 tokens per session scan C 7aecddc3f31d
ui-reviewer is an agent published in the GitHub repository alexjbarnes/cockpit (14 stars, last pushed 4d ago), licensed Apache-2.0. It adds 94 tokens to every session and 2,599 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it C with 2 findings (recursive force delete, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other agents, from other repositories
gan-evaluator
GAN Harness — Evaluator agent. Tests the live running application via Playwright, scores against rubric, and provides actionable feedback to the Generator.
ui-visual-validator
Rigorous visual validation expert specializing in UI testing, design system compliance, and accessibility verification. Masters screenshot analysis, visual regression testing, and component validation. Use PROACTIVELY to verify UI modifications have achieved their intended goals through comprehensive visual analysis.
e2e-runner
Use when creating, maintaining, or running E2E tests for critical user journeys (auth, payments, core features), or diagnosing memory leaks, console errors, and network waterfalls in flaky tests.
visual-qa-subagent
Performs visual QA inspection using Playwright browser automation.
e2e-runner
An end-to-end testing role for checking complete user journeys through a website or application. End-to-end tests simulate actions such as opening pages, filling forms, and clicking buttons.
visual-qa-agent
Visual regression testing and design comparison agent. Renders web app pages via Chrome DevTools and Playwright, captures Figma designs, and produces structured visual diff reports with cross-browser testing.