Manifold MCP: SEO MCP and Marketing Research Tools

Manifold is a hosted SEO MCP and marketing data service for people who already work with an AI assistant. It connects research tools to Claude, ChatGPT, Cursor and other compatible clients, so the assistant can retrieve evidence for an SEO audit, campaign brief or prospect shortlist.

In plain English: what does Manifold do?

Think of Manifold as a research desk your assistant can call. You provide a keyword, website, brand or company; Manifold retrieves the relevant data; your assistant explains it and turns it into a plan. The service supplies the data connection, while the assistant supplies the conversation and analysis.

Its marketing tools are read-only. They can inspect rankings or public posts, but they do not publish an article, send outreach, change an ad campaign or edit your website. It suits SEO practitioners, growth teams and agencies who want evidence inside an existing chat workflow.

Build an SEO brief from keywords, competitors and site checks

Keyword research returns ideas with metrics, while SERP tools show what Google ranks for a query. Domain overviews, ranked keywords, keyword gaps and backlink data help you identify competitors and examine opportunities before commissioning content.

Give the assistant the target country and language rather than accepting the default US/English market. Ask it to connect each recommendation to a returned keyword, result URL or audit finding. Most SEO datasets come from DataForSEO; a search-volume estimate is a provider metric, not your own conversion data.

For technical work, check one page or run a site crawl. A crawl returns a task identifier that the assistant must follow up with get_task. Google is the source for these general search datasets; the separate connected-account tools cover Bing webmaster data.

Check whether AI answers mention and cite your brand

A practical AI visibility baseline starts by checking whether answer engines can access the site, identifying relevant prompts and running those prompts against selected engines. Results expose answer text, brand mentions and cited sources, so you can distinguish being mentioned from receiving a link.

Supported answer sources include ChatGPT, Claude, Gemini, Perplexity and Google AI answers. ChatGPT and Gemini results come from their consumer apps; Claude and Perplexity use model APIs with web search. Answers vary between runs, so this is a dated observation rather than a promise of stable placement.

Put Search Console and GA4 beside competitor estimates

Authorize the accounts you own in the app's Connections area. List Search Console or Bing properties before requesting queries, pages, clicks or indexing information. For GA4, list properties and request sessions and key events by channel for the same reporting period.

These own-account calls cost zero Manifold credits but have workspace rate limits and require the relevant account permissions. They help check whether a promising keyword or AI-search initiative is associated with actual traffic and conversions.

Research public conversations, ads and B2B prospects

Social tools cover public information from Reddit, TikTok, Instagram, Facebook, YouTube, LinkedIn and X. Depending on the platform and tool, you can research profiles, posts, comments or transcripts. Public ad-library tools examine Meta, TikTok, LinkedIn and Google ads; they expose only fields the library makes available.

Company and people searches support prospect research by attributes such as industry, location or role. Finding a work email is a separate operation from a people search and can return no match. Treat these results as research inputs; Manifold does not send the campaign or guarantee contact accuracy.

Connect Manifold and run a bounded first query

1. Open your client's custom MCP connector settings and add https://mcp.manifoldmcp.com/mcp with the name Manifold.
2. Complete the browser sign-in with Google or email and authorize the client. In Claude, enable the connector in the chat's tools menu.
3. Start small: “Use Manifold to return the top 10 Google results for my seed keyword in my target country and language; show URLs and the retrieval date.”
4. Ask for a free dry_run estimate before larger jobs, set max_credits and check the result's provider, credits_charged, cached and data_as_of metadata.

The client setup guides cover the supported connection paths. OAuth uses a personal workspace; scripts or a team workspace can use a bearer API key created in the appropriate workspace.

Understand the free trial and per-call credit costs

Verified on 14 October 2026: a new workspace receives 250 free credits once, with no card required and no monthly renewal. The displayed Pro starting tier is US$20/month for 2,500 credits; Max starts at US$250/month for 42,000. Higher credit tiers are available, and applicable taxes are additional.

Ten Google results cost 1 credit; 100 keyword ideas with metrics cost 10. Calls have different formulas, so 250 credits does not mean 250 identical research jobs. Plan credits reset monthly; non-expiring top-ups are available on Pro and Max. Cached results are cheaper, and an identical result already paid for by your workspace is free until expiry. Check current plans and the credit rules before budgeting.

Check freshness, data handling and reuse limits

A response may be cached for anything from an hour to 90 days, depending on the data. AI-answer runs and technical crawls are not cached. Check the retrieval timestamp; null means the provider has no data, while zero is a measured value.

The privacy policy describes usage logs, including query targets, and 30-day retention for slower task parameters and results. Returned data also enters the connected assistant, whose own privacy rules apply. The service terms restrict dataset resale or public redistribution and unlawful use of contact data. A read-only connection still requires a decision about which accounts and information to share.

Official resources and developer workflows

• Tool documentation: parameters, results and supported research operations.
• Official marketing skills: reusable multi-step workflows for compatible agents.
• Account and billing support: help with access, credits and unexpected tool results.

Frequently asked questions

What can an SEO MCP connection actually do here?

It gives your assistant keyword, SERP, domain, backlink and audit data, alongside AI visibility and broader marketing research tools. Publishing and implementing recommendations remain separate steps.

Which assistants can connect?

Official guides cover Claude, Claude Code, ChatGPT, Codex, Cursor, Gemini CLI, Grok and OpenCode. Your chosen client must support the required connection and authorization method; its own plan conditions are separate.

Is Manifold free to keep using?

The trial is 250 credits once. Continued paid-data usage requires Pro or Max after that balance is spent; paid plans can buy top-ups. Own-account Search Console, Bing and GA4 tools are documented as zero-credit operations with rate limits.