agentos.guide › blog
GEO MCP: What It Is, The Real Repos, And How It Fits An Agent OS Workflow
A GEO MCP is an MCP server that gives your AI agent generative engine optimisation tools, so it can audit pages for AI search visibility by itself.
MCP means Model Context Protocol, and GEO means getting your business named when someone asks Google AI, Gemini or ChatGPT a question.
I read the README of every real GEO MCP I could find on GitHub today, and they're all small community projects with between zero and three stars.
This quick start shows the ranked options, how to add one to Claude Code or Hermes, and where it fits in the Search Console loop I run from my Agent OS.
If you'd like help first, book a free one-to-one GEO, AEO AI audit and get a custom game plan.
The best GEO MCP options ranked
Agent OS is my own product, so this is my own list, and the table puts it first for a reason I'll explain below.
| Rank | Option | What it is | Best for |
|---|---|---|---|
| 1 | Agent OS (my own, inside the AI Profit Boardroom) | My agent dashboard running the Search Console and SEO skill workflow from my video under one shared memory. | Anyone who wants the full GEO loop, from data to published page. |
| 2 | Hermes Agent on its own | The free, open-source agent running the same process without my dashboard. | People who want the free version of the same workflow. |
| 3 | tmolavi/mcp-geo-server | A Python MCP server for three-pillar audits across technical SEO, entity AEO and GEO readiness. | Deep page audits with an evidence label on every finding. |
| 4 | jpurnell/GeoSEOMCP | A Swift MCP server with 29 offline scoring tools for text you supply. | Mac users who want repeatable scores. |
| 5 | SEOSiri-Official/aeo-geo-mcp | A local-first Python MCP server with 10 AEO and GEO tools. | Quick llm.txt and schema density checks. |
| 6 | henu-wang/geoscore-mcp | A JavaScript wrapper for the hosted GEOScore scanning service. | Hosted scans, if the service is reachable. |
The four repos are ranked by how much their READMEs cover, how fresh the code is and how clearly they explain their limits.
I haven't run those four myself, so what follows about them comes from their own READMEs and GitHub data I pulled on 19 October 2026.
What a GEO MCP is in one minute
MCP stands for Model Context Protocol, an open standard for connecting AI apps to outside tools, and the MCP introduction explains it well.
GEO stands for generative engine optimisation, which is getting your business named when someone asks Google AI, Gemini or ChatGPT a question.
A GEO MCP is a server of GEO tools that your agent can call, such as AI crawler checks, schema validation and passage citability scores.
In an Agent OS, that just means one more toolbox your agents can reach from the shared workspace.
Why Agent OS sits above every GEO MCP
A GEO MCP gives your agent a set of checks, but the hard part of GEO is deciding what to publish and then publishing it.
In my video above, the workflow I use to get named in AI answers doesn't use a GEO MCP at all.
It uses an agent, my Google Search Console data and an SEO skill.
I run it from my Agent OS, or from Hermes Agent directly, because it's the same process either way.
Hermes Agent is free and open source, and I use OpenCode as the free brain.
When I recorded, DeepSeek V4.1 Flash was free inside OpenCode.
The Agent OS runs a crew of AI agents, such as Hermes and Claude, under one shared memory, and it's only available inside the AI Profit Boardroom rather than on GitHub.
Quick start: the Agent OS GEO loop, with an optional GEO MCP
This is how a GEO MCP slots into the loop I already run, as an optional audit step.
| Step | Who does it | What happens |
|---|---|---|
| 1. Export | You | You export Google Search Console performance data to a CSV file, which is free. |
| 2. Analyse | Agent | The agent reads the CSV and suggests keyword and topic ideas to rank inside AI search engines, based on what's working. |
| 3. Audit (optional) | Agent plus a GEO MCP | A reviewed GEO MCP checks the target pages for crawler access, schema and citability. |
| 4. Write | Agent with the SEO skill | The agent drafts the page from a real case-study angle. |
| 5. Fact-check | You | You check every claim, title, description, schema field and alt text by hand. |
| 6. Publish | Agent | The agent publishes to your site once you approve the page. |
In step two, Hermes builds a report in the workspace that shows clicks, brand share of clicks, non-brand click-through rate and the month's clicks.
It also lists what's working, such as comparison articles, Agent OS topics, the free AI community and Hermes topics, then gives a verdict on the best pages.
I'm honest in the video that this is Google organic data, because Google and the AI engines don't report AI traffic well yet.
My SEO skill is a checklist of roughly 100 points, and it writes everything from a case-study angle.
My demo prompt asks Hermes to create an SEO article using the SEO skill for the keyword "GEO examples" and publish it to aiprofitboardroom.com.
📞 Want a strategist to read your Search Console data with you?
Book a free one-to-one GEO, AEO AI audit and my team will look at your site, your competitors and the keywords you're closest to winning.
🔥 Want the exact AI SEO setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
The four GEO MCP repos, as Agent OS tools
Here's each repo with the job I'd give it inside an Agent OS, based on its README.
| Repo | Language | Stars on 19 Oct 2026 | Last code push | Agent OS job it could do |
|---|---|---|---|---|
| tmolavi/mcp-geo-server | Python | 3 | 18 September 2026 | A weekly audit of your top pages, with findings labelled from fact to hypothesis. |
| jpurnell/GeoSEOMCP | Swift | 0 | 17 October 2026 | Scoring draft passages for citability before an agent publishes them. |
| SEOSiri-Official/aeo-geo-mcp | Python | 0 | 2 October 2026 | A quick llm.txt and schema density pass. |
| henu-wang/geoscore-mcp | JavaScript | 1 | 5 March 2026 | A hosted scan, but only once the service loads again. |
tmolavi/mcp-geo-server
The README describes a Python adapter that exposes the SAGE audit engine to agents like Claude Desktop, Cursor and Antigravity.
It audits a live URL or raw HTML across technical SEO, entity AEO and GEO readiness, including AI crawler rules for GPTBot, ClaudeBot and PerplexityBot.
It labels each finding from E0 for facts to E5 for hypotheses, and it says its Citation Survival Proxy is a diagnostic rather than a citation probability.
The README installs it from PyPI, but I couldn't find the package on the public PyPI index on 19 October 2026, so you may need the repo source.
jpurnell/GeoSEOMCP
The README lists 29 tools across citability, content analysis, crawler access, schema, technical SEO, brand presence, llms.txt and composite scores.
It needs Swift 6.2 or newer and macOS 14 or newer, and it installs with a clone, swift build and swift run geoseo-mcp-server.
The README says no tool fetches a URL, so your agent passes the page text in and the scores stay offline and repeatable.
SEOSiri-Official/aeo-geo-mcp
The README calls it a local-first server for AEO, GEO, llm.txt auditing and content stickiness analytics, by Momenul Ahmad of SEOSiri.
Its 10 tools include an llm.txt audit, a 0 to 100 GEO readiness score, answer-card extraction and a schema density check.
henu-wang/geoscore-mcp
The README describes a server for the hosted GEOScore platform, with eight tools, where the deep scan, the fix generator and the citation check need an API key.
The code hasn't been pushed since 5 March 2026, geoscoreai.com returned an error when I tried it on 19 October 2026, and I couldn't find the package on npm that day.
How to add a GEO MCP to Claude Code
These commands come from Anthropic's Claude Code MCP docs.
# remote server with an HTTPS URL claude mcp add --transport http <name> <url> # local server, launch command from the server's README after the double dash claude mcp add <name> -- <command> [args...] # check it connected claude mcp list claude mcp get <name>
The docs save a server to local scope by default, and you can add --scope project to share it through a .mcp.json file or --scope user to load it everywhere.
Inside Claude Code you can also type /mcp to see each server and its status.
How to add a GEO MCP to Hermes Agent
These steps come from the Hermes Agent MCP docs, which say MCP support ships with the standard install.
# ~/.hermes/config.yaml
mcp_servers:
geo:
command: "<command from the server's README>"
args: ["<args from the server's README>"]
tools:
include: [<only the read-only tools you reviewed>]
A remote server takes a url instead of a command, and Hermes discovers the tools when you start it with hermes chat.
The tools: include: list means only the tools you reviewed get registered.
If a server won't connect, the docs say hermes mcp test reports what the server actually answered.
After a change, /reload-mcp picks it up without a whole new session, and hermes import-agent claude-code can migrate servers you set up in Claude Code.
🔥 Want the exact AI SEO setup I run inside my Agent OS?
Inside the AI Profit Boardroom you get the installable Agent OS, step-by-step video tutorials, four coaching calls a week and 3,400+ members building real automations.
Review every GEO MCP before your agents touch it
Anthropic's docs say to verify you trust each server before connecting it, and they warn that servers that fetch outside content can expose you to prompt injection.
The Hermes docs say you should read what a catalogue entry installs and runs before you install it.
So I read the source, list which tools read and which write, whitelist only read-only tools and test on a throwaway project first.
I'd also keep any GEO MCP I'm still testing well away from the agent that publishes to my sites.
The weekly skill file I'd give an agent
Once a GEO MCP has passed that review, this is the read-only skill file I'd add to an Agent OS.
GEO MCP AUDIT SKILL (weekly, read-only) 1. Load the latest Search Console CSV summary and the target page list from shared memory. 2. Pick the five pages with the most impressions for topics you want to be named for in AI answers. 3. Call only the whitelisted GEO MCP audit tools on those pages. 4. List the top three fixes per page, and mark each one as technical, content or schema. 5. Ignore llms.txt scores unless the user asks, because Google says AI features need no AI text files. 6. Save the findings to the workspace and report what changed since last week. NEVER: write files, change pages, publish, call tools that are not whitelisted, or treat a proxy score as a promise.
Line five is there because Google's AI features page says you don't need AI text files or special schema to appear in AI Overviews or AI Mode.
📞 Not sure which pages to audit first?
Claim a free AEO SEO strategy session and leave with a plan built around the keywords you're closest to winning.
Where Google's 1 October 2026 guidance fits
Google updated its page on using generative AI content on 1 October 2026.
It says it's "critical to manually factcheck and review all AI-generated content" before publishing.
Google says that review covers titles, meta descriptions, structured data and image alt text as well.
So step five in the loop is a human job, even when a GEO MCP generated the schema.
What this loop has done for me
These are my own results from my own screens in the video, not a promise.
For "best geo expert", my content shows in the Google AI answer at the top and ranks first organically.
In AI Mode, asking for the best GEO experts to follow in 2026 lists me first and links to my sites.
One of my sites went from zero to more than 1,300 clicks a day, and two other sites are on the same path.
AI Mode describes me as widely recognised as a top practitioner because of public testing and deployment, which is why the loop writes from real case studies.
What's inside the AI Profit Boardroom for this
Inside the AI Profit Boardroom you get the Agent OS, my SEO skill checklist and the AI SEO automation section.
You also get Hermes trainings that include Hermes with Claude MCP, four coaching calls a week and a member map with 3,400+ members.
Limits you should know
Every GEO MCP I found is a small community project with between zero and three stars.
A GEO score is a proxy, and no tool can tell you what an AI answer will say tomorrow.
Nobody can guarantee you'll be named in an AI answer, and that includes me.
If you'd rather not guess where to start, book a free strategy session and see why competitors appear in AI answers more often.
Where to go next on agentos.guide
My GEO examples guide shows four real AI answer wins from my own screens.
My free GEO audit guide helps you check your site before you build anything.
My Hermes plugins marketplace guide covers the same review-before-install habit for plugins.
Also On Our Network
FAQ
What is a GEO MCP?
A GEO MCP is a Model Context Protocol server that gives an AI agent generative engine optimisation tools, such as AI crawler checks, schema validation and passage citability scores.
Which GEO MCP servers exist?
On 19 October 2026 I found jpurnell/GeoSEOMCP, tmolavi/mcp-geo-server, SEOSiri-Official/aeo-geo-mcp and henu-wang/geoscore-mcp on GitHub. All four are small community projects with between zero and three stars.
How does a GEO MCP fit an Agent OS?
It is an optional audit step in the loop. The agent analyses your Search Console export, a reviewed GEO MCP checks the target pages, the agent writes from a case study, you fact-check and the agent publishes.
How do I add a GEO MCP to Claude Code?
Anthropic's docs use claude mcp add with a URL for a remote server, or a launch command after a double dash for a local server. You check it with claude mcp list or /mcp.
How do I add a GEO MCP to Hermes?
The Hermes docs say you add it under mcp_servers in ~/.hermes/config.yaml with a command and args or a url. You can limit its tools with tools include.
Do I need a GEO MCP to do GEO?
No. My own workflow uses an agent, a Search Console CSV and an SEO skill with no GEO MCP. Review any MCP server's code and permissions before you install it.
Final word on GEO MCP
Start with your Search Console export and the loop above, then add a reviewed GEO MCP only where it saves your agents real work.
When you want a second pair of eyes, book a free AEO SEO strategy session and get a plan to rank in Google and AI search.
That's how I'd put a GEO MCP to work inside an Agent OS.
About Julian
I'm Julian Goldie, an AI entrepreneur, SEO expert and the founder of the AI Profit Boardroom, which has 3,400+ members.
I help business owners scale with AI agents, automation and SEO.
My YouTube channel has 400,000+ subscribers, and I run Goldie Agency, a seven-figure SEO agency.
Related guides on agentos.guide
Get the whole Agent OS
The installable system behind every guide on this site — dashboard, agents, memory and pipelines — updated as new tools land, with 3,400+ members and four coaching calls a week.
Join the AI Profit Boardroom →📺 Video notes + links to the tools 👉
🎥 Learn how I make these videos 👉