Token导航 LogoToken导航TokenDH.com
研究检索操作浏览器clawhub未标认证来源可访问clear审计提醒

tap点击

Agent Skill

tap 用于处理浏览器自动化、网页检查和页面信息提取,适合在 OpenClaw 中需要让 Agent 打开页面、读取网页或验证前端流程时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

3,672

周安装

150

GitHub Stars

公开资料未说明

下载量

1,176
OpenClaw

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

MIT-0

最后核验

2026-05-01

来源状态

来源可访问

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:tap(点击)
来源仓库:https://github.com/leonting1010/tap
安装命令:
openclaw skills install tap
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 OpenClaw 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

ClawHubOpenClaw
openclaw skills install tap

简介

AI 浏览器自动化协议 — 为 41 个站点运行预构建的技能,或创建新的技能。 MCP 原生、确定性、运行时零 AI。

SKILL.md

name
tap
description
AI browser automation protocol — run pre-built skills for 41 sites, or forge new ones. MCP native, deterministic, zero AI at runtime.
version
0.1.2
metadata
openclaw
requires
bins
install
formula
LeonTing1010/tap/tap
bins
[tap]
emoji
🪶
homepage
https://github.com/LeonTing1010/tap

Tap — The Interface Protocol for AI Agents

Tap gives you deterministic browser automation. Instead of burning tokens on every click, forge a script once and run it forever — zero AI at runtime.

How It Works

Tap exposes MCP tools in 6 categories:

Run Pre-Built Skills (zero AI, instant results)

Use tap.list to see all 81 available skills, then tap.run to execute:

tap.run({ site: "github", name: "trending" })        → trending repos
tap.run({ site: "hackernews", name: "hot" })          → top HN stories
tap.run({ site: "zhihu", name: "hot" })               → Zhihu trending
tap.run({ site: "xiaohongshu", name: "search", args: { keyword: "AI" } })

These run in under 1 second, cost $0, and return structured data every time.

81 skills across 41 sites: X/Twitter, Reddit, GitHub, YouTube, Bilibili, Zhihu, Xiaohongshu, Weibo, Medium, arXiv, Hacker News, Product Hunt, Bluesky, Steam, CoinGecko, and more.

Forge New Skills (AI creates, then never needed again)

When you need a site that doesn't have a pre-built skill:

  1. Inspectforge.inspect({ url: "https://example.com" }) analyzes the page structure and available data sources
  2. Verifyforge.verify({ url: "...", expression: "..." }) tests the extraction logic live
  3. Saveforge.save({ site: "example", name: "data" }) persists the skill

After saving, tap.run({ site: "example", name: "data" }) works forever. No AI needed.

Direct Browser Control

Operate the browser via the page API for one-off interactions:

  • page.nav({ url }) — navigate to a page
  • page.click({ target }) — click by selector or visible text
  • page.type({ selector, text }) — type into input fields
  • page.find({ query }) — find elements by text
  • page.screenshot() — capture the current page
  • page.scroll, page.hover, page.pressKey, page.select
  • page.fetch({ url }) — make API requests in the page context

Inspect & Debug

  • inspect.dom — page DOM structure
  • inspect.a11y — accessibility tree
  • inspect.page — page metadata and state
  • inspect.resources — loaded resources

Tab Management

  • tab.list — all open tabs
  • tab.new({ url }) — open new tab
  • tab.close({ tabId }) — close tab

Setup

1. Install Tap

Download the latest binary from GitHub Releases and add to PATH.

Or build from source (requires Deno):

git clone https://github.com/LeonTing1010/tap && cd tap
deno compile --allow-read --allow-write --allow-net --allow-env --allow-run --output tap src/cli.ts

2. Install Chrome Extension

Download tap-extension.zip from the latest release, unzip, load as unpacked extension in chrome://extensions/.

3. Install Community Skills (optional)

tap install    # 81 skills across 41 sites

4. Add MCP Server

Add to your OpenClaw MCP configuration:

{
  "mcpServers": {
    "tap": {
      "command": "tap",
      "args": ["mcp"]
    }
  }
}

Common Workflows

Research: aggregate trending across platforms

1. tap.run github/trending
2. tap.run hackernews/hot
3. tap.run reddit/hot
→ Cross-reference results for emerging topics

Monitor: track topics across sites

1. tap.run x/search { keyword: "AI agents" }
2. tap.run zhihu/search { keyword: "AI agents" }
3. tap.run xiaohongshu/search { keyword: "AI agents" }
→ Compare discussion across platforms

Publish: cross-post content

1. tap.run x/post { content: "..." }
2. tap.run xiaohongshu/publish { title: "...", content: "..." }
3. tap.run telegraph/publish { title: "...", content: "..." }

Forge: create a skill for any new site

1. forge.inspect { url: "https://newsite.com" }
2. forge.verify { url: "...", expression: "..." }
3. forge.save { site: "newsite", name: "data" }
4. tap.run newsite/data  ← works forever, zero AI

Security & Trust

Provenance. Tap is open source (AGPL-3.0) at github.com/LeonTing1010/tap. All release binaries are built via GitHub Actions — verify by checking the CI workflow.

Chrome extension permissions. The extension requires debugger permission to send CDP commands to the active tab. It does NOT request <all_urls>, cookies, or webRequest in its manifest. The extension only activates when Tap is explicitly invoked.

Community skills. tap install clones scripts from tap-skills. All scripts are plain .tap.js files (readable JavaScript) — review before running. User-forged taps are stored locally in ~/.tap/taps/.

Scope of access. Tap operates on the active browser tab when invoked. It does not run in the background, does not access tabs you haven't navigated to, and does not persist any data beyond ~/.tap/.

Recommendation. Review the source code (~1,800 lines) and extension manifest before installing.

适合场景

01

OpenClaw 用户查找和安装 Skill 时

02

用户想查找某类 Agent Skill 时

03

需要根据任务场景推荐可安装能力包时

04

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

补充不同宿主或平台的使用分布数据

能力 5

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

OpenClaw

87.04%
按下载量换算1,024

安全审计

VirusTotal

可疑

ClawScan

可疑

Static analysis

通过

权限和风险

操作浏览器

该 Skill 可能涉及浏览器控制能力,使用时可能读取或操作网页内容,需要在受控环境中确认权限边界。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills