mcpField
L
Link Preview Cache API

Link Preview Cache API

Remote
github.com/cmondilloBrowser Automation0Updated 9w ago
Repository

About Link Preview Cache API

Link Preview Cache API provides paid URL metadata extraction through a shared cache, at about one-third the price of link-preview-api. It is aimed at AI agents needing page previews.

Key features

  1. 01URL title and metadata extraction
  2. 02Shared-cache responses
  3. 03About one-third the price of link-preview-api
  4. 04USDC micropayments via x402

Use cases

  1. 01Retrieve page previews for agent workflows
  2. 02Reduce costs for repeated link lookups

MCP support

The portfolio offers some services as MCP tools, but the input does not state whether Link Preview Cache API itself has an MCP interface.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://link-preview-cache-api.onrender.com/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

Tools

Details

From the project README

tools

A portfolio of small, single-purpose paid services sold directly to AI agents — no human checkout flow, no dashboard login, no subscription. An agent calls the API, gets an HTTP 402 Payment Required quote, pays a few cents in USDC, and gets its answer. Fully automatic on both sides.

Each service lives in its own self-contained folder under projects/ and follows the same pattern:

  • One narrow problem, solved well. No platforms, no suites — a single job an agent needs done.
  • x402 for payment. HTTP-native, stablecoin micropayments, no merchant account or KYC required on the seller side — just a wallet address. This is the emerging standard for agent-to-API payments (backed by Coinbase, adopted by AWS Bedrock AgentCore, Cloudflare, and others as of 2026).
  • Discoverable by agents, not just humans — each API declares an x402 Bazaar discovery schema so agents can find and call it without a human ever reading the docs.
  • A sample automation client included with every project, showing an agent discovering, paying for, and consuming the service end-to-end.
  • Where it adds real reach, also sold as an MCP tool (api/app/mcp_server.py) — the same product, the same price, a second interface for coding agents that discover tools over MCP rather than raw HTTP.

Projects

ProjectWhat it sellsStatus
link-preview-apiPay-per-call URL metadata (Open Graph/title/description/image) extractionLive on Base mainnet, earning real USDC: https://link-preview-api-z4nf.onrender.com
content-moderation-apiPay-per-call profanity/explicit-content detection with redaction, sold over both HTTP and MCPLive on Base mainnet, earning real USDC: https://content-moderation-api-hhy1.onrender.com
link-preview-cache-apiSame product as link-preview-api, ~⅓ the price — backed by a shared cache. First instance of a general "x402 cache" patternLive on Base mainnet, earning real USDC: https://link-preview-cache-api.onrender.com

Repo conventions

  • Every project is fully self-contained in projects/<project-slug>/: its own code, tests, Dockerfile, deploy config, and README. Nothing product-specific is shared between projects.
  • Every project ships with:
    • api/ — the paid service itself
    • automation-client/ — a sample agent that pays for and consumes it
    • deploy/ — ready-to-use deploy configs (Render/Fly/etc.)
    • README.md — the business case, architecture, and exact steps to take it from "runs on my machine" to "earning money"
    • LISTING.md — the marketplace/directory listing copy, ready to publish
  • New projects get added the same way: a clean new folder, same skeleton, linked in the table above.
  • One deliberate exception to "nothing shared": scripts/ holds portfolio-level publishing tooling — not product code, reused across every project the same way (e.g. submitting a new tool to x402-list.com's API). See PUBLISHING.md for the full playbook: what's genuinely automatic, what's scripted, and what stays a deliberate manual step and why.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.