docsLABSCO SUMMARY
A vendor data API with an MCP front door — no browser state of your own.
Everything runs on Xquik's side against your API key, so there are no proxies, cookies or X logins to manage, and the filters run before you are billed rather than after. The tradeoff is that coverage, schema and the bill are all theirs: deleted, protected or restricted content can stay out of reach, some bounded reply reads come back as 424 with partial rows and a diagnostic, and the docs do not yet tie every deduplicated row to the final credit deduction — so run the estimate route before a large job and treat what it returns as a forecast, not an invoice.
What it is
Xquik's hosted MCP server at https://xquik.com/mcp, sitting in front of the company's own REST API for X (Twitter) data. Xquik is an independent third-party service, not affiliated with X Corp. The MCP surface is a route-discovery surface rather than one tool per endpoint: the exposed names are docs, search and execute, and execute takes a path plus a query — a xquik.request against /api/v1/x/tweets/search carrying q, language, minLikes and limit. The live OpenAPI behind it documented 128 HTTP operations when the README was checked against it on 2026-08-23. MCP responses come back with normalized snake_case fields; binary support downloads go through REST instead. Current clients negotiate MCP 2026-07-28 through server/discover. The same repository is an Agent Plugins v1 package — x-developer v2.6.7 bundles the x-twitter-scraper and xquik-social-research Skills alongside the hosted server.
What you get
- Tweet reads — lookup, batch lookup, search, timelines, replies, quotes, threads, likes, reposts and media, returning
id,text,createdAt,likeCount,retweetCount,replyCount,quoteCount,viewCount,bookmarkCountand anauthorblock - Profile reads — lookup, batch lookup, search, followers, following, relationships, account details and availability
- Other X surfaces — Lists, communities, trends, Spaces, articles, bookmarks, notifications and supported feeds
- Filters applied server-side, before delivery — Tweet filters cover authors, recipients, mentions, languages, dates, media, engagement, verification, phrases, words, hashtags, cashtags, URLs, lists, places, replies, reposts and quotes; profile filters cover follower counts, following counts, Post counts, account age, verification, websites, locations, bios and usernames
- Bulk extraction across 23 extraction types, with multi-target jobs, result caps,
dedupeAcrossTargetsanddedupeMode, cursor pages and file exports - A cost estimate you can run first — the estimate route returns
allowed,estimatedResults,creditsRequired,creditsAvailableandsource, and consumes no credits - Exports in JSON, CSV, Markdown, PDF, TXT and XLSX
- Account and keyword monitors — active monitors check every second, with stored events and signed webhooks included; webhook deliveries carry
X-Xquik-Timestamp,X-Xquik-NonceandX-Xquik-Signature, and event cursors let you recover after downtime - Write actions on a connected X account in the same contract — posts, replies, deletes, likes, reposts, follows, DMs, media, profiles and communities
- Support-desk calls — create tickets, send messages, upload and download attachments, poll processing status
Requirements
An XQUIK_API_KEY, created in the Xquik dashboard; an MCP client authenticates with OAuth or that API key. You do not need an official X developer account, and you do not need to connect an X account at all for Tweet, profile, search, follower, timeline, reply, quote, repost or media scraping. Private reads and X account actions are the exception — those need an X account connected first. Never send an X password, cookie, session export or 2FA code to Xquik or to an agent; the Skill does not install packages, run local bridge commands, write local files, browse local networks or load remote code. Billing is per delivered result and runs on credits: 1 credit per returned Tweet on search, 1 credit per result for most extraction rows, 5 credits per article extraction row, and no credits for estimates or for reading and exporting a stored job. The PAYG rate is $0.00015 per credit, the first PAYG funding amount is $10, and unused credits carry over. Each active monitor costs 21 credits per hour. A low balance can stop results at the affordable count; zero affordable results return 402 insufficient_credits.
Setup effort
Add the hosted endpoint, then sign in — paste it into your client, then approve the OAuth sign-in
USE CASES
Collecting a filtered Tweet set on a topic
Ask for the narrow set up front — language, a minimum like count, replies and reposts and quotes excluded — then follow next_cursor while has_next_page is true. A filtered page can come back empty and still have another page after it.
Finding the right route before spending anything
The API is wide — 128 documented HTTP operations — and the MCP surface is built for looking a route up rather than guessing at one.
Pricing a bulk pull before you run it
Send the extraction body to the estimate route first: it answers with allowed, estimatedResults, creditsRequired and creditsAvailable and spends nothing. Send the same body to create the job once the numbers look right.
Exporting a dataset someone else can open
Create the extraction, store the returned id, poll the job, page with nextCursor while hasMore is true, and export only after it completes — JSON, CSV, Markdown, PDF, TXT or XLSX.
Watching an account or keyword over time
Create a monitor with the event types you want and take the results from stored events or a signed webhook. Verify X-Xquik-Timestamp, X-Xquik-Nonce and X-Xquik-Signature on delivery, and keep the HMAC secret from creation time.
Posting, replying or sending a DM from your own X account
Same contract as the reads, but the X account has to be connected first. Confirm the account, target, payload and estimated cost before the call, use idempotency keys, and poll an ambiguous write before retrying it.
3 tools
Listed in project docs. Your client may show a different list.
searchexecuteBEFORE YOU INSTALL
What to know first
Tools you get
3 tools named in the project's documentation.
Requires
- an MCP client that can reach a remote endpoint
- An Xquik account — billing is per delivered result, on credits
INSTALLATION
Connect your client
Server setup
https://xquik.com/mcpAdd this hosted endpoint to your MCP client, then sign in.
MCP Clients
Any MCP client — remote
{
"mcpServers": {
"x-twitter-scraper": {
"url": "https://xquik.com/mcp"
}
}
}Clients that support OAuth open a browser to sign in to Xquik the first time they connect, so no key goes in the configuration. Where a client cannot do OAuth the API-key fallback is client-specific — an older Codex release, for example, takes bearer_token_env_var = "XQUIK_API_KEY" in its server entry.
Codex
codex plugin marketplace add Xquik-dev/x-twitter-scraper
codex plugin add x-twitter-scraper@x-twitter-scraperClients that support OAuth open a browser to sign in to Xquik the first time they connect, so no key goes in the configuration. Where a client cannot do OAuth the API-key fallback is client-specific — an older Codex release, for example, takes bearer_token_env_var = "XQUIK_API_KEY" in its server entry.
Claude Code
claude mcp add --transport http xquik https://xquik.com/mcpThen run /mcp, select xquik and authenticate in the browser.
Clients that support OAuth open a browser to sign in to Xquik the first time they connect, so no key goes in the configuration. Where a client cannot do OAuth the API-key fallback is client-specific — an older Codex release, for example, takes bearer_token_env_var = "XQUIK_API_KEY" in its server entry.