Menu
Drive the Boardex UI against the mock runner in a headless browser to verify a change end-to-end.
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
npx skills add https://github.com/ankayca/boardex --skill verifyThe command stays on one line. Scroll horizontally to inspect it before copying.
Prefer a local copy? Download the files currently available to SkillsMP.
Based on SOC occupation classification
Showing SKILL.md
| name | verify |
| description | Drive the Boardex UI against the mock runner in a headless browser to verify a change end-to-end. |
Ports 4319 (mock runner default), 4321, 5173, 5199 are often occupied by leftover
dev processes — pick fresh ports and check with ss -tln first. Never
pkill -f "tsx src/index.ts" to clean up: the pattern matches every mock runner
on the machine, including ones you don't own. Kill by PID.
# mock runner (SPEED scales fixture delays; acceptance runs use SPEED=5)
cd tools/mock-runner && SPEED=5 PORT=4333 npx tsx src/index.ts
# UI, pointed at it
cd apps/ui && VITE_RUNNER_URL=http://localhost:4333 npx vite --port 5333 --strictPort
# health checks
curl -s http://localhost:4333/health # {"ok":true,...,"runnerKind":"mock"}
No browser tooling in the repo. Install Playwright in the session scratchpad
(NOT the repo): npm init -y && npm install playwright && npx playwright install chromium (skip --with-deps; sudo is unavailable and the WSL box already has
the libs). Viewport ≥1280px wide for the three-zone workspace grid; 1100px to
check the right rail stacking under center.
Useful handles (accessible roles, stable):
header span[data-kind="status"],
complementary "Board context", step rows are <li> with an expand button named
by step title, log tabs role=tab (Agent/Build/Flash/Serial/RTT), panes
role=log named " — log", lines under [data-index] (virtualized
— only ~30 rows render; short logs render fully).GET /runs/{id}/events?afterSeq=0, find unresolved approval.requested, then
POST /runs/{id}/approvals/{aid} {"status":"approved"}.The full fixture at SPEED=5 completes in ~80s with prompt approvals. Collect
page.on('console') errors — the §7.2/§7.3 done-criteria include "no console
errors". Mid-run reload is the standard replay check: completed steps' log
content must be byte-identical before/after.