skim-mcp

Skim is the complete web toolkit for AI agents—not just a simple page reader. Get a free card key at skim402.com/pricing; wallet/x402 is optional.

skim-mcp is the official Model Context Protocol server for Skim. It turns public pages into clean, agent-ready Markdown—typically ~4x smaller than raw HTML in Skim's measured comparisons—and exposes the full toolkit: clean reads, batch reads, JavaScript/browser rendering, structured extraction, crawling, PDFs, page-change watches, and curated Signals.

Default path: a card-plan API key. Get a free sk402_ key at skim402.com (pricing). Paste it as SKIM_API_KEY. A crypto wallet is not required.

Optional path: x402 wallet pay. If you prefer pay-per-call in USDC on Base, you can set SKIM_WALLET_PRIVATE_KEY instead. Ignore this unless you want it.

Tools in this package: read_url, read_urls, extract_url, crawl_url, read_pdf, watch_urls, check_watch, poll_signal.

Try before you integrate: use the Playground for a quick read or the Workbench to test batches, extraction, crawl, PDFs, watches, and Signals.

Quickstart (60 seconds)

Card API key (recommended)

1. Get a free key at skim402.com — it starts with sk402_. Card required at signup; 1,000 reads/month on the free plan. Details: skim402.com/pricing.

2. Add this to your MCP client. Claude Desktop — ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows). Cursor — ~/.cursor/mcp.json (or Settings → MCP). Same JSON works in Cline, Continue, Zed, and other MCP clients:

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

3. Restart the client and ask it to read a URL:

Read https://en.wikipedia.org/wiki/HTTP_402 and summarize it.

Wallet / x402 (optional)

Only if you prefer pay-per-call ($0.002 USDC on Base) instead of a card key. You can skip this entire block.

{
  "mcpServers": {
    "skim": {
      "command": "npx",
      "args": ["-y", "skim-mcp"],
      "env": {
        "SKIM_WALLET_PRIVATE_KEY": "0xYOUR_BASE_WALLET_PRIVATE_KEY"
      }
    }
  }
}

Fund a dedicated Base wallet with a small USDC balance ($1 ≈ 500 reads). Setup guide: https://skim402.com/wallet.

Use a fresh wallet, not your personal one. This private key lives in a plaintext config file on your machine — treat it like a hot wallet for $0.002 tolls, not a savings account.

Related MCP server: Base Intel Search

FAQ

Do I need an API key?
Yes for the default path. Get a free sk402_ key at https://skim402.com. A wallet is optional.

Is this wallet-only? Do I need crypto?
No. Card API key is the default. x402 wallet pay is optional.

What tools can my agent call?
read_url, read_urls (batch), extract_url, crawl_url, read_pdf, watch_urls, check_watch, and poll_signal.

Try it without an agent

With a card key:

curl -H 'Authorization: Bearer sk402_your_key_here' \
  'https://skim402.com/api/t/read?url=https://en.wikipedia.org/wiki/HTTP_402'

Wallet / x402 challenge (optional — returns HTTP 402 if you send no payment):

curl -i -X POST https://skim402.com/api/v1/read \
  -H 'content-type: application/json' \
  -d '{"url":"https://en.wikipedia.org/wiki/HTTP_402"}'

The tools

read_url

Reads any URL and returns clean Markdown with a YAML frontmatter block.

Input: { "url": "https://example.com/article" }

Routes: GET /api/t/read?url= (API key) · POST /api/v1/read (wallet)

Output:

---
title: Example article
byline: Jane Doe
published: 2025-01-15
lang: en
excerpt: A short summary...
---

# Example article

The cleaned article body in Markdown...

read_urls

Batch-read 1–10 URLs in one call. Optional stripLinks / stripImages.

Input: { "urls": ["https://a.example", "https://b.example"] }

Routes: POST /api/t/read/batch · POST /api/v1/read/batch

extract_url

Structured JSON from a page. Pass a JSON Schema, or a preset: article, product, job, review, event, table.

Input: { "url": "https://example.com/product", "preset": "product" }

Routes: POST /api/t/extract · POST /api/v1/extract

Presets are sent as schemas on those extract routes.

crawl_url

Crawl a site (origin or start URL) and return clean Markdown per page. Discovers sitemap.xml / robots.txt sitemaps plus same-origin links. Cap 25 pages. 1 credit per successful page; failed pages are not charged.

Input: { "url": "https://example.com", "maxPages": 10 }

Optional: stripLinks, stripImages. Bare hosts like example.com are treated as https://example.com.

Route: POST /api/t/crawl (API key). No x402 /v1 twin — set SKIM_API_KEY.

Crawl https://example.com (max 10 pages) and list the page titles.

read_pdf

Fetch a public PDF URL and return clean Markdown plus an optional bookmark outline. Text comes only from the file. Image-only scans return 422 (no OCR). Files larger than 8 MB return 413. 3 credits; failed conversions are not charged.

Input: { "url": "https://example.com/paper.pdf" }

Optional: outline (default true).

Route: POST /api/t/read-pdf (API key). No x402 /v1 twin — set SKIM_API_KEY.

Read the PDF at https://example.com/paper.pdf and summarize the outline.

poll_signal

Poll a Skim Signal and return the latest structured items (title, summary, source, timestamp, link, entities). 2 credits per successful poll; failed polls are refunded.

Requires SKIM_API_KEY. Wallet-only configs get a clear error.

Input: { "slug": "ai-news", "limit": 20 }

Optional filters (pass only those the feed documents): forms (sec-filings, campaign-finance), categories (deals), fields (research), states (film-incentives), committees (campaign-finance).

Slugs: ai-news, sec-filings, crypto-news, macro, security, regulations, courts, recalls, deals, launches, trending, research, energy, entertainment, studio-jobs, campaign-finance, film-incentives. Use x402 for the ecosystem feed.

Routes (API key): GET /api/t/signal/{slug}/latest?limit= · GET /api/t/feeds/x402/latest?limit= (x402 is not /signal/x402)

curl -H 'Authorization: Bearer sk402_your_key_here' \
  'https://skim402.com/api/t/signal/ai-news/latest?limit=20'

Poll the ai-news signal for the latest 20 items.

watch_urls / check_watch

Register 1–20 URLs, then poll for content diffs. watch_id is a secret.

Input: { "urls": ["https://competitor.com/pricing"], "note": "pricing" } then { "watch_id": "w_…" } (optional status_only: true)

Routes (API key): POST /api/t/watch · GET /api/t/watch/diff?id= · GET /api/t/watch/status?id=

Routes (wallet): POST /api/v2/watch · GET /api/v2/watch/diff?id= · GET /api/v2/watch/status?id=

Example agent prompts

Read https://en.wikipedia.org/wiki/HTTP_402 and summarize it.

Read these three pages and compare their pricing: https://a.example/pricing https://b.example/pricing https://c.example/pricing

Extract the product name, price, and availability from https://example.com/products/notebook as JSON.

Crawl https://example.com (max 10 pages) and list the page titles.

Read the PDF at https://example.com/paper.pdf and summarize it.

Poll the ai-news signal for the latest 20 items.

Watch https://competitor.com/pricing and https://competitor.com/changelog, then check the watch for changes.

Configuration

How it actually works

Card lane (default):

your agent ──► skim-mcp ──► GET https://skim402.com/api/t/read?url=…
                                 Authorization: Bearer sk402_...
                                        │
                                        ▼
                                200 OK + clean Markdown

Wallet lane (x402, optional):

your agent ──► skim-mcp ──► POST https://skim402.com/api/v1/read
                   ▲                       │
                   │                       ▼
                   │              402 Payment Required
                   │                  (x402 challenge)
                   │                       │
                   ▼                       │
   x402-fetch signs EIP-3009 ◄────────────┘
   USDC transfer authorization
                   │
                   ▼
        retry POST with X-PAYMENT header
                   │
                   ▼
   Skim verifies + settles via Coinbase CDP
                   │
                   ▼
        200 OK + clean Markdown

End-to-end latency is typically 1.5–2 seconds including settlement. Wallet-lane private keys never leave your machine — they only sign authorizations locally.

Security

• Card key: stored in your MCP client's JSON config. Anyone with read access to that file can use your key. Keep your key on the free plan or a small-balance plan.

• Wallet key (only if you use the optional wallet path): lives in the same JSON config in plaintext. Anyone with access to your home directory can drain the wallet. Keep it funded with only as much USDC as you're willing to spend in a runaway loop.

• No outbound telemetry from this package. skim-mcp only talks to skim402.com (or whatever you set as SKIM_API_URL). No analytics, no error reporting, no phone-home.

Troubleshooting

"No tool named read_url shows up in Claude/Cursor." Restart the client fully (quit, don't just reload the window). MCP servers are spawned at client startup. If still missing, run npx skim-mcp directly in a terminal — if you get a stack trace, it's likely Node < 18.

"401 Unauthorized" with card key. Your SKIM_API_KEY is invalid or expired. Get a fresh one at skim402.com/pricing.

"402 Payment Required loops forever" (wallet lane). Your wallet is out of USDC on Base mainnet. Check the balance on BaseScan, top up if needed.

"USDC is on Ethereum, not Base." This is the most common funding mistake. USDC on Ethereum mainnet does not work — Skim only accepts USDC on Base. See https://skim402.com/wallet for the bridging walkthrough.

Links

• Skim website — https://skim402.com

• Pricing & free key — https://skim402.com/pricing

• Wallet setup guide — https://skim402.com/wallet

• API docs — https://skim402.com/docs

• x402 protocol — https://x402.org

• Model Context Protocol — https://modelcontextprotocol.io

• GitHub — https://github.com/JessieJanie/skim402

License

MIT