One MCP server.
36 platforms of data.

It's a Model Context Protocol server at https://mcp.fetchlayer.dev that exposes every FetchLayer platform — social feeds, app stores, ad libraries, review sites, maker communities and storefronts — as tools your AI agent or IDE can call directly. Instead of writing wrapper code that calls our REST API and parses the response, your agent calls the tool directly — search Reddit, pull App Store reviews, scrape a YouTube video's comments — and gets structured JSON back inline, in the same conversation.

Any MCP-compatible client: Claude Desktop, Claude Code, Cursor, Windsurf, VS Code (via an MCP-compatible extension), Hermes, Kilocode, and OpenClaw are all supported today, with setup guides for each. If your tool speaks MCP over stdio or HTTP, it can connect — the protocol is the same regardless of client, only the config file location and format change.

No. MCP calls draw from the same API credits as direct REST calls — one tool call is one request, at the same $1.99 per 1,000 requests pay-as-you-go rate, or covered by your subscription plan if you have one. There is no separate MCP tier, no markup, and no additional key to manage: the same Bearer token that authenticates your REST calls authenticates your MCP connection.

Yes. Rather than shipping a separate MCP server per platform, FetchLayer runs a single server that exposes every platform's tools together, namespaced by platform (fetchlayer.reddit.search, fetchlayer.youtube.comments, and so on). You connect it once and your agent has access to whichever platforms your plan includes — no juggling multiple server URLs or config blocks as you add platforms, and nothing to re-configure when a new one ships.

Anything the REST API can do: search and read Reddit posts and comments, pull a subreddit's feed, read a TikTok account's videos, a hashtag feed or the comments on a video, fetch public App Store or Google Play reviews for an app, extract YouTube video comments, look up Google Maps or Trustpilot reviews for a business, pull every ad a competitor is running across the Facebook, Google, TikTok and LinkedIn ad libraries, read Glassdoor salaries, follow a Product Hunt launch, walk a Shopify or WooCommerce store's catalogue, check live Amazon prices, search Google, Bing and DuckDuckGo and check where a site ranks, or find out whether ChatGPT mentions and cites a brand in its answers — all as callable tools your agent invokes mid-conversation instead of you copy-pasting API responses in by hand.

No. Setup is a config block — your API key and the server URL, pasted into whichever file your client uses for MCP servers. Each client-specific guide below has the exact file path and a copy-paste config. There's no SDK to install and no server to run yourself; the server is hosted and maintained by FetchLayer.

Yes, freely — they're the same underlying service. Use the REST API from your backend for scheduled jobs or pipelines, and the MCP server from your IDE or agent for interactive, conversational lookups. Same API key, same usage counted against the same plan.