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 ZhanlinCui/Agent-Skills-Hunter --skill web-accessgit clone --depth 1 https://github.com/ZhanlinCui/Agent-Skills-HunterWrote 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/zhanlincui/agent-skills-hunter/web-access)<a href="https://agentmods.dev/skills/zhanlincui/agent-skills-hunter/web-access"><img src="https://agentmods.dev/badge/skills/zhanlincui/agent-skills-hunter/web-access/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/zhanlincui/agent-skills-hunter/web-access"><img src="https://agentmods.dev/badge/skills/zhanlincui/agent-skills-hunter/web-access.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 | $0.00089 | $0.02808 |
| Opus 5 | $0.00044 | $0.01404 |
| Sonnet 5 | $0.00018 | $0.00562 |
| Haiku 4.5 | $0.00009 | $0.00281 |
Grade A, and why
web-access 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 10d 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 — 200 lines — stays where its author put it; the contents beside it link to each section on GitHub.
web-access Skill
首次安装
用户首次使用时，执行以下流程：
Step 1：运行环境探测
bash ~/.claude/skills/web-access/scripts/check-deps.sh
Step 2：AI 根据输出处理缺失依赖
探测脚本只报告事实，安装决策由 AI 完成。缺什么装什么，Chrome 缺失时提示用户手动下载（无法自动安装）。
Step 3：安装完成后，向用户说明以下内容
web-access 已就绪。凡是联网的需求直接说就行，我会自动选最合适的方式：
- 只需要搜索结果 → 直接搜，最快
- 需要看完整页面 → 抓取页面内容，不启动浏览器
- 需要登录或动态页面 → 自动启动浏览器，登录一次后持久保存
Windows 用户需要 Git Bash 环境（安装 Git for Windows 即可）。
浏览哲学
像人一样浏览，不像机器人一样执行程序。
人类浏览网页时不会在开始前列出完整步骤，而是带着目标进入，边看边判断，遇到阻碍就解决，发现内容不够就深入——全程围绕「我要拿到什么」做决策。这个 skill 的所有行为都应遵循这个逻辑。
三个核心判断：
① 我需要什么？ — 任务驱动，先想清楚目标信息的性质，再选最轻且能直达的方式。不要用重型工具做轻量任务，也不要用轻量工具面对它覆盖不到的内容。
② 够了吗？ — 拿到的信息能完成任务，就是够了。不过度采集，不为了"完整"而浪费代价。大概了解一个视频，几帧就够；理解一篇文章，读文字就够；不需要全页截图去做能用 accessibility tree 完成的事。
③ 遇到阻碍怎么办？ — 在层内解决，不退回，不打扰用户。弹窗、登录墙、广告、加载失败——像人一样判断这个阻碍是否真的挡住了目标内容：挡住了就处理，没挡住就绕过去继续。只有在确认无法自行解决时才告知用户。
信息获取通道选择
- 先评估任务，再选通道：根据「目标信息的性质、什么工具能直接拿到」决定起点，选最轻且能直达的方案。
- 确保信息的真实性，一手信息优于二手信息：搜索引擎和聚合平台是信息发现入口。当多次搜索尝试后没有质的改进时，升级到更根本的获取方式：定位一手来源（官网、官方平台、原始页面）。
| 场景 | 通道 |
|---|---|
| 只需搜索摘要或关键词结果，或需要发现信息来源 | WebSearch |
| URL 已知，静态公开页面 | WebFetch |
| 社交媒体、内容平台（微信公众号、微博、小红书、X/Twitter 等） | 浏览器 CDP（直接，跳过 WebFetch） |
| 需要动态内容、登录态、交互操作，或需要像人一样在浏览器内自由导航探索 | 浏览器 CDP |
浏览器 CDP 不要求 URL 已知——可从任意入口出发，通过页面内搜索、点击、跳转等方式找到目标内容。
WebFetch 请求时加 header Accept: text/markdown, text/html，支持该协议的网站直接返回 Markdown，省约 80% token。失败（空内容 / 403 / JS 渲染）时升级到浏览器层。
降级禁止：进入更重的通道后，不得回头用轻量工具完成同一目标——等同于重走已知不通的路。浏览器层遇到阻碍应在层内解决（如处理登录），而不是绕回。唯一例外：浏览器操作中衍生的新子目标，可重新选择通道。
进入浏览器层后，区分任务性质：
- 操作型（导航、填表、点击）：用 accessibility tree 感知界面，无法识别时才截图辅助
- 内容型（读帖子、看资讯、分析页面）：accessibility tree 读文字结构，同时判断图片是否承载核心信息——是则提取图片 URL 定向读取
图片判断：社交媒体、图文博客、截图类内容，默认图片有价值，主动去取；工具类、导航类页面，默认 accessibility tree 够用。
浏览器 CDP 模式
启动
bash ~/.claude/skills/web-access/scripts/ensure-browser.sh
Browser ready on port 9222→ 脚本自己启动的，状态可信，直接用（任务结束后关闭）already running→ 检测到残留进程，状态不可信，必须验证：运行agent-browser --cdp 9222 open about:blank，成功则可用；失败则执行 close 后重新 ensure（任务结束后不关闭）ERROR或 agent-browser 无响应 → 执行bash ~/.claude/skills/web-access/scripts/close-browser.sh后重新运行
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 10d ago First seen · 200 lines · 89 tokens per session scan A 00eb45019ca8
web-access is a skill published in the GitHub repository ZhanlinCui/Agent-Skills-Hunter (186 stars, last pushed 6mo ago), licensed MIT. It adds 89 tokens to every session and 2,808 once invoked, about $0.0004 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
opencli-sitemap-author
Use when creating or maintaining OpenCLI site sitemaps: agent-facing navigation, page-state, action, workflow, API-reference, pitfall, and fallback knowledge for a website. Use after browser exploration discovers durable site context, when a sitemap is stale, or when promoting local site knowledge into the repo.
debug-optimize-lcp
Guides debugging and optimizing Largest Contentful Paint (LCP) using Chrome DevTools MCP tools. Use this skill whenever the user asks about LCP performance, slow page loads, Core Web Vitals optimization, or wants to understand why their page's main content takes too long to appear. Also use when the user mentions…
interactive-login
How to complete browser/interactive logins (aws / gh / glab / gcloud). The platform backgrounds the login poller so it survives the human's browser round-trip — and when that does NOT work.
pinchtab-mcp
Use this skill when a task requires browser automation through PinchTab's MCP server connected to a remote browser instance. Covers navigation, element interaction, data extraction, form filling, multi-step flows, and session management via MCP tools.
azure-messaging-webpubsub-java
Build real-time web applications with Azure Web PubSub SDK for Java. Use when implementing WebSocket-based messaging, live updates, chat applications, or server-to-client push notifications.
google-safe-browsing
Prevent and fix Google Safe Browsing "Dangerous site" flags. Use when launching a public web app, buying/picking a domain, building a login or signup page, or when any site shows a red "Dangerous site" / "Deceptive site" warning in Chrome, Brave, Safari, Firefox, or Edge. Triggers on "dangerous site", "deceptive…