Taprun

Name: Taprun
Author: LeonTing1010

@LeonTing1010

关于 Taprun

Your AI agent scrapes X today, gets broken HTML tomorrow. Taprun compiles the site once into a deterministic program, then `tap doctor` monitors health contracts + structural diffs to catch silent breakage before your data goes stale. 140+ pre-built skills across 68+ sites. Works

#browser-automation #mcp #web-scraping #deterministic #agent-tools #claude-code #cursor #windsurf

基本信息

分类

其他

传输方式

stdio

发布者

LeonTing1010

提交者

DATA D

配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

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

工具

未检测到工具

工具是从 README 中自动提取的。维护者可以在 ## Tools 标题下列出工具,即可填充这部分内容。

概览

What is Taprun?

Taprun compiles AI understanding into deterministic programs called .tap.js scripts, then monitors them for breakage. It replaces ad-hoc web scraping with repeatable, zero-cost extractions and alerts when the source changes. The server is built for developers and AI agents that need reliable, always-fresh data from live websites without silent staleness.

How to use Taprun?

Install via npx -y @taprun/cli (zero-install) or curl -fsSL https://taprun.dev/install.sh | sh (permanent). Connect to an MCP host by adding { "mcpServers": { "tap": { "command": "tap", "args": ["mcp", "start"] } } } to your client config. Run skills with tap <source>/<endpoint>, e.g. tap github/trending or tap hackernews/hot.

Key features of Taprun

forge: AI inspects a site once and emits a .tap.js program
tap run: program executes instantly with zero LLM cost
tap doctor: health contracts and structural state diffs catch breakage
heal: AI reads diagnostics and patches the program only when needed
140+ community skills in one binary with zero dependencies

Use cases of Taprun

Monitor trending repositories on GitHub without manual scraping
Fetch live headlines from HackerNews or Weibo with deterministic output
Detect and auto-repair extraction programs when a website’s HTML shifts
Replace recurring API calls or LLM-based scraping with a local, cost-free script

FAQ from Taprun

How does Taprun differ from regular web scraping?

Taprun compiles AI understanding into a deterministic .tap.js program that runs at zero cost. It monitors the program with health contracts and structural diffs, so you are alerted before stale data enters your pipeline — not after.

What dependencies does Taprun require?

None. Taprun is distributed as a single binary and works out of the box. You can run it via npx or the shell installer with no additional setup.

Where does scraped data live?

The README does not specify a storage location. Data is fetched live upon execution of a .tap.js program; it is not persisted in a backend unless configured by the user.

What happens when a website changes?

The tap doctor command detects structural shifts using health contracts and state diffs. If a break is found, the heal command invokes AI to diagnose the issue and patch the program — only when needed.

How do I connect Taprun to my MCP host?

Add the following to your MCP host configuration: { "mcpServers": { "tap": { "command": "tap", "args": ["mcp", "start"] } } }. No special authentication or transport setup is required.