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 agentmods add skills/migoxlab/webqa-agent/button-checknpx skills add MigoXLab/webqa-agent --skill button-checkgit clone --depth 1 https://github.com/MigoXLab/webqa-agentWhat 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 | $0.00015 | $0.02185 |
| Opus 5 | $0.00008 | $0.01092 |
| Sonnet 5 | $0.00003 | $0.00437 |
| Haiku 4.5 | $0.00002 | $0.00218 |
Grade A, and why
button-check 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 yesterday.
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 — 189 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Button Check Skill
Systematically test every interactive element on the page:
- Clickables (button / link / tab / checkbox / radio / switch / menuitem) — click and verify
- Inputs (textbox / searchbox / combobox / slider) — fill with sample data and verify
After each test, return to the baseline page so the next element is tested from the same starting state.
Important: Before each tool call, output a short one-line description of what you are about to do (e.g. "Filling the search box with 'hello'"). This helps generate readable step-by-step reports.
When to Use
- Task mentions "遍历所有按钮/链接/输入", "comprehensive UI testing", or "regression testing"
- You need to verify no broken links / dead buttons / unresponsive inputs exist
- Smoke testing a page before a release
Skip for single-element tests — use click or fill directly instead.
Phase 1: Collect Interactive Elements
Call take_snapshot to get the accessibility tree. Group elements into two buckets by role:
Clickable bucket — driven by click:
button,link,tabcheckbox,radio,switchmenuitem
Input bucket — driven by fill (or type_text / press_key for keyboard-only widgets):
textbox,searchboxcombobox(treat as click-to-open if no editable affordance is visible, then fill if a textbox appears)slider(fill with a numeric mid-range value)
Each element has a uid (e.g. uid=1_5), a role, and a label. Record {uid, role, label, bucket} for every element.
Cap at 50 elements total. If more exist, prioritize:
- Inputs (form fields) — these often gate downstream behavior
buttonandlink(core interactive)tab,menuitem(navigation)- Other ARIA roles
Snapshot misses some elements? Icon-only buttons without ARIA attributes are invisible to take_snapshot. Mention in the report instead of erroring.
Phase 2: Baseline
Record the baseline state before testing:
- Call
take_screenshotwithfullPage: trueto capture the entire page - Note the current URL — this is the baseline URL you MUST return to after every test that navigates away
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.
- yesterday First seen · 189 lines · 15 tokens per session scan A b4fbc02a05b4
button-check is a skill published in the GitHub repository MigoXLab/webqa-agent (228 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 15 tokens to every session and 2,185 once invoked, about $0.0001 per session on Opus 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 2026-08-30.
Other skills, from other repositories
printed-curtain-render
Generate a browser-based p5.js curtain simulation that weaves user-supplied PNG, JPEG, or WebP artwork and optional text into individually moving strands. Use when users ask for an interactive curtain or thread simulation, artwork woven into cloth or ropes, image-bearing fabric motion, 可交互线帘/幕布, 图片织入布料, printed…
threejs-earth-render
Clone or update https://github.com/vibe-motion/threejs-earth and render the Three.js Earth route animation with Puppeteer frame capture. Use when users ask for 三维地球航线动画, Three.js Earth, 地球飞线, globe route animation, or exporting an Earth GIF/MP4/PNG sequence.
find-docs
Use globally installed ctx7 for API/library docs when local docs or CLI help are insufficient. Trigger for API syntax, config, setup, migrations, version-sensitive, or post-cutoff questions. Do not use for web research, local files, browser automation, or installs.
tavily-web
Use Tavily CLI for web search, URL extraction, site mapping, and cited research. Trigger when the user needs current web info, asks to search/read a page, provides URLs, or wants a sourced report. Do not use for browser automation, local docs, or Tavily setup.
dingtalk_channel_connect
Use a headed browser to automatically complete DingTalk channel integration for QwenPaw. Applicable when the user mentions DingTalk, developer console, Client ID, Client Secret, bot, Stream mode, binding or configuring a channel. Supports pausing when a login page is detected and resuming after the user logs in.
browser_cdp
当用户明确希望连接到已运行的 Chrome 浏览器、扫描本地 CDP 端口、显式指定 cdpport，或让多个 agent / 工具共享同一个浏览器时，使用本 skill。browser 默认不开放调试端口；仅当用户明确希望其他本地工具附加时才显式传入 cdpport。.