SEO Inside Google Antigravity: Live Search Data

You are three files deep in a build inside Google Antigravity. The Agent Manager is working through a task, you are drafting copy for a new pricing page, and a question surfaces. Is anyone actually searching for the phrase you just dropped into the headline, or are you about to optimize a page for a keyword nobody types? The reflex is to break flow, open a browser, log into a $129 dashboard, run one lookup, and lose your place by the time you get back.

There is a cleaner answer now. Antigravity is Google’s agentic development platform, a VS Code-based editor with an Agent Manager that runs tasks on your behalf, and it speaks the Model Context Protocol. That is the same plug standard that lets an editor talk to GitHub, Postgres, and your file system. Point it at a live SEO data source and you can ask that keyword question in the agent panel, get a real volume and difficulty number back in seconds, and never leave the file you were editing.

Antigravity is brand new, so wiring SEO into it is still a first-mover setup, and there is one config detail that trips up almost everyone. This guide covers how to connect the data, the gotcha coming from Cursor, what you can actually ask once it is live, and the honest limits of doing SEO this way. I build solo, I write my own copy, and I run this workflow every day, so this is how it actually works, not a feature list.

Why do SEO inside Antigravity

The case is not that Antigravity is suddenly an SEO tool. It is that the cost of asking an SEO question should be close to zero, and right now it is not.

Most builders who do their own SEO use a thin slice of what a dashboard offers, and only now and then. You want to know whether a keyword is winnable before you write the post. You want a fast read on a competitor’s traffic. You want to check that the headline you just wrote matches what people actually search. None of those need a 40-tab research session. They need one number, quickly, without a context switch out of the editor.

The Model Context Protocol, or MCP, is what makes this work. It is an open standard, introduced by Anthropic in late 2024, that lets AI agents call external tools and data sources through one common interface. Antigravity is built on VS Code and ships with an Agent Manager that can run tasks and call tools for you, so once you connect a server the agent reaches its tools mid-task without you spelling out which one to use. An SEO data source is just another tool it can pick up.

The payoff is flow. When the answer to “is this keyword soft enough to rank” lives in the same window as your code and your copy, you ask it. When it lives behind a login two tabs away, you guess instead, and guessing is how you end up with a page tuned for a phrase nobody searches. If you want the longer argument for why asking the agent beats opening a dashboard, the SEO MCP server explainer makes the full case.

How do you connect SEO data to Antigravity?

Antigravity does not make you hunt for a config file on disk. You open it from the UI, which is the reliable route because the path has moved between builds.

Open Settings, go to the Customizations tab, choose Manage MCP Servers, then View raw config. That opens mcp_config.json in the editor. On disk the file lives under your home .gemini folder, and Antigravity 2.0 shares one MCP config between the editor and the CLI so a single entry lights up both, but open it from the UI rather than guessing the path.

A hosted server that exposes one remote URL is the least work. Here is the shape for a remote SEO server like ContextBolt SEO, which hands you a single URL on signup. Add it under the mcpServers key.

{
  "mcpServers": {
    "contextbolt-seo": {
      "serverUrl": "https://seo.contextbolt.app/mcp/YOUR_TOKEN"
    }
  }
}

Here is the detail that trips up everyone coming from another editor. The key is serverUrl, not url. Antigravity inherits its config style from Windsurf, where the field is serverUrl, and this is the single most common reason the tools never appear. It is worth memorizing, because every neighboring tool names the same field differently. Cursor and VS Code use url. The Gemini CLI uses httpUrl. Codex uses url in a TOML file. Antigravity uses serverUrl. Paste a block copied from a Cursor guide and it will silently do nothing.

Because your token sits in the URL path, you do not need a headers block, though Antigravity supports one for servers that expect a bearer token. That token is personal, so treat the URL like a password and keep mcp_config.json out of any public repo. Once the block is saved, hit Refresh in the MCP Servers panel, then ask the agent “what SEO tools do you have?” to confirm the tools loaded. If they are missing, the fix is almost always the serverUrl key or a Refresh you skipped.

What can you actually ask once it’s connected?

This is where it stops being a config exercise and starts earning its keep. You type questions, the agent picks the tool, the data comes back in the panel. A few of the moves I run most often.

• Keyword reality checks: “What keywords should my new pricing page target?” The agent runs keyword research and hands back related terms with volume, difficulty, and intent, so you plan the page while you build it.
• Difficulty plus SERP in one ask: “How hard is ‘gemini mcp server’ to rank for, and who’s in the top 10?” It chains a difficulty score with the live page-one results, so you know the odds and the competition before you commit to a format.
• Domain teardowns: “What does supabase.com rank for, and who else competes?” It runs ranked keywords and then competitor keywords, giving you a content map plus the rivals chasing the same terms.
• Backlink gap mining: “Who links to my main competitor but not to me?” It returns a spam-scored prospect list, a ready-made outreach shortlist. Backlink calls are the pricier ones, so this is a deliberate spend.
• Difficulty triage: “Of these five keywords, which can a new domain realistically rank for?” One prompt sorts a shortlist into write-now and wait-for-authority.
• Audit and fix in one run: “Audit my site and fix the top three issues.” The Agent Manager grades your pages worst first, rewrites the top offenders, and re-audits to show the gain, without a second prompt.

The thread through all of these is plain language and chaining. You are not learning a menu or remembering which report lives where. You ask the way you would ask a colleague, and because these are MCP tools the agent strings several together on its own. “Find five low-difficulty keywords for my features page, check who ranks for each, and note the gaps” is a single prompt that fires several tools and writes the whole session to disk.

Where ContextBolt SEO fits

ContextBolt SEO is the one I make, and it exists for exactly this workflow. It is a hosted MCP server that wraps Ahrefs-grade data and hands you one URL. You paste that URL into Antigravity’s mcp_config.json, hit Refresh, and you have nine tools in the agent. Six research tools, keyword research, keyword difficulty, SERP overview, domain overview, ranked keywords, and competitor keywords, cost one lookup each. Three backlink tools, backlink overview, referring domains, and competitor gap, cost three each. The plan is $35 a month for 1,000 lookups, which is a quarter of the entry price of the big suites, and there is no account to provision or deposit to fund. You subscribe, you get a URL, you work. The free Google Search Console tools and the automatic cross-session memory never spend a lookup.

Two things make it fit Antigravity specifically rather than just any MCP client. First, Antigravity’s agent has file access, so every lookup mirrors to a ./seo-findings/ folder in your workspace as markdown, one clean file per keyword or domain, plus an index. Your SEO research lives in the repo next to the feature it informed. You can grep it, commit it, diff it next month, or open it in Obsidian. Second, that memory compounds. Ask about the same keyword a week later and the answer leads with what changed since you last looked, with no extra prompt. The step-by-step version lives on the Antigravity integration page, and the full tour of every tool is in the ContextBolt SEO guide.

There is one more move, and the Agent Manager is what makes it land. Say “audit my site and fix the top three” and ContextBolt SEO grades the mechanical health of your pages against a strict ruleset, title and meta lengths, a single H1, thin content, broken links, canonical tags, and hands back its own 0-100 score worst first. Because the Agent Manager edits files on its own, it then rewrites those pages and re-audits, leading with the delta, health 80 to 91, in one run. Ahrefs reports the problems. ContextBolt hands them to the thing that can fix them. Find it, then fix it.

The honest framing the product holds to applies here too. This is Ahrefs-grade data, not the same numbers as Ahrefs. It sits on DataForSEO estimates, which are decision-useful and directionally accurate, the same class of data inside many dashboards you already trust. For deciding what to build and what to write, that is more than enough.

The honest limits

Doing SEO inside Antigravity is not a magic replacement for everything a suite does, and pretending otherwise would be the kind of overclaim this blog tries to avoid.

You do not get a visual dashboard. If you think in charts and like clicking around a keyword explorer, the agent panel will feel sparse at first. You do not get scheduled rank tracking. An MCP server answers when you ask, it does not watch your positions every day and email you when something moves, so if daily rank monitoring is core to your job, keep a tool that does it. Lookups are metered, so a monthly budget means you spend it on real questions rather than idle curiosity, which in practice nudges you toward the checks that actually change a decision. And the agent can be confidently wrong about strategy even when the underlying numbers are right, so treat its SEO advice the way you treat its code, useful and fast and in need of a human read before it ships.

What you gain in return is the thing that matters most when you build solo and time is the scarce resource. The answer arrives where you already are, in the seconds you would otherwise spend context switching, at a price that does not make you wince. Antigravity is new enough that most people have not wired live data into it yet, so getting this in place now means the next time a keyword question lands mid-build, you just ask. The phrase you were nervous about might be wide open, and now you can find out without ever leaving the file.