# JessieJanie/skim402 [Health: Active]

**Category:** 🔎 Search & Data Extraction  
**Repository:** https://github.com/JessieJanie/skim402  
**GitHub Stars:** 2  
**npm Downloads (last month):** 442  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/jessiejanie-skim402

## Description
Convert any URL into clean, agent-ready Markdown with structured metadata (title, byline, published date, language). Pay-per-use via x402 (USDC on Base) — no signup, no API keys. Single tool readurl. Install via npx -y skim-mcp.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "skim402": {
    "command": "npx",
    "args": ["-y","skim-mcp"],
    "env": {
      "SKIM_API_KEY": "",
      "SKIM_WALLET_PRIVATE_KEY": ""
    }
  }
}
```

**Requires environment variables:** `SKIM_API_KEY`, `SKIM_WALLET_PRIVATE_KEY` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What JessieJanie/skim402 MCP server does

JessieJanie/skim402 MCP server connects an MCP client to Skim's web-content APIs. Its primary read operation accepts a public URL and returns cleaned Markdown with a YAML-style metadata block containing fields such as title, byline, publication date, language, and excerpt. This is intended for agents that need page content in a smaller, more usable form than raw HTML.

The package covers more than single-page reading. Agents can submit several URLs at once, request JSON that follows a supplied schema, crawl pages within one site, convert public PDFs to Markdown, register URLs for later change checks, and poll predefined Signals feeds for structured items. The available tools are `read_url`, `read_urls`, `extract_url`, `crawl_url`, `read_pdf`, `watch_urls`, `check_watch`, and `poll_signal`.

## How it works

The MCP server is launched as a local Node package with `npx`. MCP client configuration passes requests to Skim's hosted service rather than performing all extraction locally. The default authentication method is a `SKIM_API_KEY` value from Skim's card-plan service. The README also documents an optional `SKIM_WALLET_PRIVATE_KEY` route for pay-per-call USDC transactions on Base.

Single and batch reads return cleaned page content. Extraction accepts either a JSON Schema or a named preset such as article, product, job, review, event, or table. Crawling follows same-origin links and available sitemap information, with a maximum of 25 pages. Watch operations first register one to 20 URLs and then use the returned secret watch ID to request status or content differences.

## Setup and configuration

Install the JessieJanie/skim402 MCP server by adding the following command to an MCP client configuration:

```json
{
  "mcpServers": {
    "skim": {
      "command": "npx",
      "args": ["-y", "skim-mcp"],
      "env": {
        "SKIM_API_KEY": "sk402_your_key_here"
      }
    }
  }
}
```

The README gives configuration locations for Claude Desktop and Cursor and says the same JSON can be used with Cline, Continue, Zed, and other MCP clients. A free card-plan key begins with `sk402_`; the documented free plan includes 1,000 reads per month. Instead of the card key, callers can configure a dedicated Base wallet through `SKIM_WALLET_PRIVATE_KEY` for optional x402 payments. The wallet path is not available for every operation: crawling, PDFs, and Signals require the API-key path, and Signals specifically requires `SKIM_API_KEY`.

## Tools and capabilities

- Read one URL as cleaned Markdown with metadata.
- Read 1–10 URLs in a batch, with options to remove links or images.
- Extract structured JSON using a schema or supported preset.
- Crawl a site and return cleaned Markdown for up to 25 pages.
- Convert a public PDF to Markdown with an optional bookmark outline.
- Track changes across registered URLs and retrieve diffs or status.
- Poll named Signals such as `ai-news`, `security`, `regulations`, and `research`.

## Limitations and notes

The JessieJanie/skim402 MCP server reads public URLs; the supplied material does not describe authenticated-site browsing. PDF conversion accepts files up to 8 MB and does not perform OCR, so image-only scans return an error. Crawling charges one credit per successful page, PDF conversion costs three credits, and successful Signal polls cost two credits; failed operations are described as not charged or refunded where applicable. The README documents a free plan as well as optional paid wallet usage, so usage limits and costs should be checked before deploying repeated crawls or polling jobs.

_Full upstream README: https://allmcps.com/mcp/jessiejanie-skim402/readme_