Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add makifbaysal/tasktrooper --skill frontend-manual-testinggit clone --depth 1 https://github.com/makifbaysal/tasktrooperWrote 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/skills/makifbaysal/tasktrooper/frontend-manual-testing)<a href="https://agentmods.dev/skills/makifbaysal/tasktrooper/frontend-manual-testing"><img src="https://agentmods.dev/badge/skills/makifbaysal/tasktrooper/frontend-manual-testing/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/skills/makifbaysal/tasktrooper/frontend-manual-testing"><img src="https://agentmods.dev/badge/skills/makifbaysal/tasktrooper/frontend-manual-testing.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 | $0.00044 | $0.01747 |
| Opus 5.5 | $0.00018 | $0.00699 |
| Sonnet 5.5 | $0.00009 | $0.00349 |
| Haiku 4.5 | $0.00004 | $0.00175 |
Grade A, and why
frontend-manual-testing scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 6d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 92 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Frontend Manual Testing
Overview
A frontend task is verified by rendering it: boot the app, drive the changed flows in a real browser, and look at what the user would see. Screenshots are mandatory evidence — a UI verdict without having looked at the screen is an untested claim.
Preconditions: case matrix recorded with record_test_cases (scenario-plan-first), covering both functional cases (per AC) and visual cases (layout, states, responsiveness); environment chosen per test-environment-selection.
Booting
- First choice
start_task_preview(test-environment-selection); only when it cannot run the project, install and start per the project's declared commands (npm ci+ dev/preview script, or the repo'sbuild_command) with the log at$QA/dev.log, never/tmp/dev.log. Point the app at the local backend you booted (backend-manual-testing) or at the stage API base URL — the API base is usually an env var (VITE_API_URLor similar); read the project's config, don't guess. - Confirm the page actually loads before scenario work: a blank page with console errors is finding #1.
- Confirm the app is talking to the real backend, not mock data (MSW/
VITE_USE_MOCKS/fixtures):requestsfrom playwright-cli (below) or the backend's own access log shows the actual calls.
Shared browser
The browser tools' session is shared with every agent running right now — there is one headless browser for the whole server. Each screenshot result prints url: and viewport WxH: confirm both are what you set before judging the image; a mismatch means another run moved it — re-navigate, re-set the viewport, retake. If the page suddenly shows a login screen, log in again.
Driving flows and taking screenshots
The browser tools are the primary way to drive the app — no local Chromium setup, no script files:
browser_navigateto the page under test.browser_wait_forthe element or text that proves the page rendered (an eternal spinner or a timeout here is finding #1 — never screenshot a page you have not waited on).browser_screenshotthe screen at all four widths of the four-width check — phone (browser_set_viewport {device:"mobile", width:360}), tablet ({device:"tablet"}), laptop ({device:"desktop", width:1024}), desktop ({device:"desktop"}) — for every changed screen's resting state and its empty/error state. Each viewport switch emulates size, touch and the matching user agent together, and answers the responsive question in words: whether the page scrolls sideways and which elements overflow. Report what it says at each width; horizontal overflow is a finding.
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.
- 6d ago Changed 11715bdf77af
- 8d ago First seen · 92 lines · 44 tokens per session scan A ae1c4b3bb2c1
frontend-manual-testing is a skill published in the GitHub repository makifbaysal/tasktrooper (113 stars, last pushed yesterday), licensed Apache-2.0. It adds 44 tokens to every session and 1,747 once invoked, about $0.0002 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 2026-10-05.
Other skills, from other repositories
qa-acceptance
Produce QA acceptance criteria and a manual validation plan for a feature change — golden path, edge cases, error states, performance limits, and explicit pass/fail evidence.
jetson-video-pipeline
Use when planning, executing, and independently validating Jetson Video Codec SDK or PyNvVideoCodec encode/decode, transcode, segmentation, container decode, AV1, or concise acceptance workflows.
proof-e2e
Build E2E test specs for critical user journeys — Playwright or Cypress, page objects, setup/teardown, CI config. Use when asked to "write E2E tests", "end-to-end testing", "browser tests", "UI tests", or "Playwright tests".
proof-strategy
Produce a test strategy for a project or feature — risk map, test type decisions, coverage targets, CI config. Use when asked to "create test strategy", "what should we test", "testing plan", or "improve test coverage".
proof-recon
Testing reconnaissance — inventory all tests, frameworks, coverage, CI integration, and assess testing maturity for project takeover. Use when asked to "understand the tests", "testing assessment", "what's tested", or "test inventory".
webapp-testing
Verify a local web app in a real browser with page, element, click, form, console and screenshot evidence. Use after a build or UI change; do not implement features or publish deployments with this skill.