Ubersuggest API: There Is No Public API. Use MCP, ChatGPT or Claude Instead

The short answer: no API, but a working MCP server

Ubersuggest has no public API and no webhooks ^[1], and no plan adds one at any price. What it offers instead is an MCP server with 42 tools, an official ChatGPT app and a Claude plugin, and all three work with a free Ubersuggest account ^[2]. That route covers analysis inside AI assistants: you ask a question in plain language and the assistant fetches the numbers. It does not cover scripts, dashboards or data pipelines, because something still has to ask the question. Know which of the two you need before you read further. Conversation works. Automation does not.

Most people who search for an Ubersuggest API, about 40 a month in the US by our measurement, want rankings in a dashboard, keyword data in a spreadsheet, or SEO data inside an AI assistant. Only the third is possible today. This guide covers that route in depth, then the manual exports that fill part of the gap, and finally the rivals worth pricing if you truly need machine-to-machine access.

What Ubersuggest's own pages say

The help article on the subject is direct: Ubersuggest "currently does not support either" APIs or webhooks ^[1]. It describes the product as a standalone application that does not send real-time notifications to outside services. The pricing FAQ says there are no API access plans available at the moment, and Enterprise/Agency at $99 a month, the top self-serve plan, has none either. Webhooks are missing too, so nothing is pushed out when a ranking drops. In-app alerts and email notifications exist, but only the admin receives project emails, so alerting into Slack or a CRM is not possible directly. A forwarding rule is the workaround.

Where the pages disagree

The API help article was updated in October 2025, before the MCP server and the ChatGPT app arrived, and it says the tool does not interact automatically with other software; the MCP, ChatGPT and Claude pages describe live connections. The AI Chat page counts 37+ tools, while the MCP and ChatGPT pages count 42. The ChatGPT launch post, updated August 3, 2026, described keyword research as the first live capability, with other tools coming in phases, yet the app page now says all 42 are available. The newer pages describe today. Even so, test the tool you need on day one.

The MCP server: the closest thing to an API

The MCP page defines Model Context Protocol as a new standard that lets AI assistants and development tools connect directly to tools and data sources, a universal plug for real-time access. Ubersuggest's server exposes 42 tools over OAuth 2.0, with no API key, and it works with Claude, Claude Code, Cursor, Codex, Windsurf and other MCP clients ^[2]. Calls draw on your account's daily reports and monthly credits, exactly like the web app. The assistant does the clicking. You do the thinking.

What the eight categories cover

The counts on the MCP page add up to 42 (our arithmetic). Domain analysis covers traffic, keywords, authority, top pages and competitors for any domain. Keyword research covers volume, difficulty, suggestions, SERP and trends, and it filters well when you ask for difficulty below a threshold. Backlink tools use the same data as the app, which says its backlinks come only from the Moz API; ask for domains that link to two rivals but not to you, and you have a prospect list. Site audit crawls still follow your plan's limits of 1,000, 5,000 or 10,000 pages per project ^[3]. Rankings still update weekly. MCP reads them faster but does not refresh them.

Sign-in by OAuth, not keys

There is no key to paste. The Claude page describes authentication as OAuth 2.0 with Dynamic Client Registration: on the first tool call, the assistant opens a browser, you sign in, and that is the whole setup. There is no environment variable or config file to keep in sync, the assistant never sees your password, and you can revoke access later. Clients that need a manual configuration use the server address from the Claude page, again with OAuth, so the endpoint alone gives nothing. That is safer than passing shared keys around a team.

Connecting Claude, ChatGPT and coding tools

In Claude on the web or desktop, open the official Ubersuggest connector in Claude's connector directory, click Connect and sign in. All 42 tools become available right away. A free account is enough: the Claude page says every tool works on a free Ubersuggest account, while noting that a free account, not no account, is needed. That makes MCP the cheapest SEO data connection for AI assistants we cover, since Ahrefs' MCP starts on its $129 Lite plan. The plugin route installs the same connector plus SEO skills, and either way takes about a minute.

Claude Code takes the full plugin with two commands, which pull in nine skills and the seo-strategist subagent; run /mcp to check that the connection is authenticated. For other editors, the MCP page gives setup instructions per client, and one command copies the skills into Cursor, Codex, Copilot, Zed, Warp and 70+ other agents that read skill folders. Only the plugin format is limited to Claude Code. The data still comes through the MCP server either way. The skills supply the method.

ChatGPT and AI Chat

Ubersuggest is an official app inside ChatGPT, on web, mobile and desktop once enabled, and it needs a free or paid ChatGPT account plus an Ubersuggest account. No plugin to install. The launch post from August 2026 calls it a first-party connection built on OpenAI's App platform and MCP. Keyword research is the one tool with an interactive widget showing a real table; elsewhere, results come back as data in the answer. AI Chat puts the same engine inside the Ubersuggest app itself, running on the same MCP server, and it is included in the account and rolling out to all users.

The Claude plugin: nine skills and a background agent

The Claude plugin installs SEO skills and the connector together, in five steps inside Claude. The nine skills are SEO action plan, project setup, keyword research, competitor analysis, site audit, content demand finder, content brief, AI visibility and SEO foundations, which loads behind the others. Each skill is a workflow, not a single call. Project setup builds a project from scratch, with competitors, AI topics and keywords to track. Site audit returns prioritized fixes. The content demand finder turns a business description into 50 content opportunities and needs no account. Briefs are free to run; generating a full article is the one paid action.

In Claude Code, the seo-strategist subagent takes the heavy analyses: full site audits with long crawl polling, several competitors at once and multi-market keyword research. It runs in the background and returns the finished report rather than the forty tool calls behind it, so you can keep working in the same session. Terminal users get the most out of the plugin for exactly this reason, and agencies will find it genuinely useful. It still spends your reports, though, from the same allowance as everything else you do that day.

Quotas: what MCP spends and what your plan changes

There is no separate API quota to buy. MCP calls draw on the same daily reports and monthly credits as the web app ^[4], and the Claude page states the rule plainly: "your plan changes how much data comes back, not which tools run". A free account gets the same tools with smaller results and the 3-a-day report limit. Individual raises that to 150 reports a day, Business to 300 and Enterprise to 900. Depth scales with price, and so does how long an assistant session can run before the day's allowance is gone.

That allowance goes quickly on Individual. A long Claude session that runs 30 tool calls uses a fifth of the day's 150 reports, so five such sessions would exhaust the daily allowance before any manual research in the app (our arithmetic). Credits are a separate budget: a full article costs 100 monthly credits, needs a paid plan, and Claude always asks before spending them. Monthly credits are 250, 500 or 1,000 by plan. Teams share the pool, too. Invited users connect their own AI client, but their searches count against the admin's daily limit, so on Enterprise a team of five gets 180 reports each (our arithmetic).

Getting data out without an API

Without an API, exports are how data leaves Ubersuggest. Paid plans export reports to PDF and CSV, free users export some reports to CSV, and the dashboard exports a full-page PDF and charts as PNG images. Exports run 2,000, 5,000 or 10,000 rows per report by plan, and to get every row you may need to scroll and load all results before clicking Export All. A monthly export of 5,000 keywords on Business is fine. A daily one is tedious.

Three other connections move data, each in one direction. Ubersuggest connects your own Google Analytics and Search Console for real traffic and keyword data, but only the admin can connect them. The WordPress plugin signs in with no API keys to copy or manage, runs Site Audit in wp-admin and applies AI fixes for titles, meta descriptions and H1s; it is the deepest integration Ubersuggest offers, because it writes to your site. The Chrome extension is only a viewer, with 40 to 900 daily searches by plan, and cannot save research.

What MCP cannot replace

MCP needs a person asking questions in an AI client. It cannot run on a schedule, feed a database or refresh a dashboard by itself. There is no way to pull rankings every Monday at 6 a.m. into a spreadsheet without someone in the loop, and agencies that build client dashboards from API data cannot connect Ubersuggest to them. The home page also confirms there is no white-label reporting. Dashboards need manual uploads, and that costs hours each month. Scrapers are no workaround either: the Terms forbid scraping or reselling data and allow account cancellation without refund for it.

We found no Looker Studio or Google Sheets connector on Ubersuggest's pages, though the home page lists n8n and Zapier among its apps and plugins. Those two are worth testing. We could not verify their scope. On security, MCP's OAuth sign-in keeps your password away from the assistant, but all customer data is processed and stored in the United States, in AWS us-east-1, with no choice of region, and Ubersuggest holds no public SOC 2 or ISO 27001 certification. Check your own policy before you connect client accounts.

When you need a real API: the rivals to price

You need a real API if you feed dashboards, sync rankings into a CRM, run scheduled jobs or resell data. Then choose a rival. Ahrefs Lite and SE Ranking Core both include an API at $129 a month, while Ubersuggest's top plan costs $99 and has none, so switching means paying at least $30 a month more (our arithmetic). Ahrefs Starter at $29 has neither API nor MCP. SE Ranking also tracks rankings daily, which makes it the most direct replacement for an agency that needs both. Semrush is the most expensive route we cover, but also the broadest suite. Our Ubersuggest alternatives guide compares them in full.

A week of SEO work without an API

Here is what a working week looks like with a person in the loop. Rankings update weekly, so Monday starts after the refresh. On Tuesday, ask Claude through the plugin for movers and losers across projects. On Wednesday, export the rank report to PDF for clients. On Thursday, run a content brief. On Friday, check AI visibility after its refresh. None of it needs an API, and none of it runs without you. Specific prompts work best, such as long-tail keywords under a difficulty threshold or volume trends over 12 months. Before you rely on the setup, test it during the trial:

• Run your three most common questions through the assistant.
• Confirm the numbers match the app.
• Check how many daily reports a typical session uses.
• Confirm your AI client is on the supported list.

If you still want a real API, request it through the Suggest a Feature option at the bottom of the left-hand menu; the help center says those requests feed the product team. No timeline is given. Ask support by email whether a public API is planned, whether the n8n and Zapier connections use the MCP server, and whether MCP will stay available on free accounts. Keep the answers in writing.

The bottom line

Ubersuggest has no API, and no plan adds one. Its MCP server, ChatGPT app and Claude plugin are the best substitute in this price range: free to connect, quick to set up and genuinely useful for analysis, as long as a person asks the questions. Watch the daily reports, because the assistant spends the same pool as the app. For automation, scheduled pulls or client dashboards, buy a rival with an API. For conversation, Ubersuggest is ready. For what each plan's reports and credits cost, see our Ubersuggest pricing guide.