Mindket
English
English简体中文
Sign inSign up
StoreMCPSkillsCalculatorLearnPriceBlog
Discover
Research308Writing59Business95Legal13Finance & accounting33Marketing & SEO172Developer1,379Productivity382Education81Companions14
Help and support
Language
English简体中文

GEO Optimizer

Made by Auriti-LabsSkillMarketing & SEO
Newversion 2 · updated
Star1291
Like

Overview

GEO Optimizer, picture 1
‹1 / 2›
GEO Optimizer, picture 2
‹2 / 2›

Auriti Labs’ open source program that scores a website from 0 to 100 on whether AI search can crawl, read and quote it, then writes the fix files. Its skill makes Claude run the checks in order and ask before it writes anything.

GEO Optimizer is an open-source project on GitHub in the Marketing & SEO category, listed by Mindket. It is free. This is version 2, last updated on . Install it from GitHub as its README says. Get Mindket’s setup notes on your Mindket account, as a zip.

At a glance

Runs in
  • Terminal: the geo program, from pip, pipx or uvx
  • Claude Code: a plugin that carries the skill and the MCP server
  • Cursor, Windsurf and other MCP clients: the geo-mcp server
  • Claude Projects, ChatGPT and Kiro: text files of advice with no program inside
Start with
pipx install geo-optimizer-skill

Platform

Claude CodeCoding agent
CursorCode editor
WindsurfCode editor
ClaudeAI chat
ChatGPTAI chat

About

What it is

GEO Optimizer is Juan Camilo Auriti’s open source program, geo, plus a skill that teaches Claude to run it. The program scores a website from 0 to 100 on whether ChatGPT, Perplexity, Gemini, Claude and Google’s AI answers can crawl it, read it and quote it. It then lists what to fix and can write the fix files. The skill is a rulebook for Claude. The program does the work.

  1. You install the geo program, then the plugin
  2. You ask Claude to audit your site
  3. Claude runs geo audit and reads the score
  4. Claude previews the fixes and waits for your yes
  5. Claude writes the files and you upload them

Who it suits

Site owners, developers and SEO people who want a repeatable AI search check. A developer can gate a build on the score. Someone without a terminal habit can ask Claude to run it. It measures readiness only. The README says plainly that it cannot guarantee any engine will cite you.

Where to use it

The geo program is the base. Most routes run it, and the text-file route carries its advice alone.

  • Terminal: the geo program, from pip, pipx or uvx
  • Claude Code: a plugin that carries the skill and the MCP server
  • Cursor, Windsurf and other MCP clients: the geo-mcp server
  • Claude Projects, ChatGPT and Kiro: text files of advice with no program inside
  • CI: a GitHub Action, or any CI that runs Python

How to set it up

Get it saves a zip of Mindket’s setup notes in your downloads folder, with a page for each route above. The program itself installs from the Python package index.

What you need first
NeedDetail
Python 3.10 or newerFor the pip routes. The README gives it as the requirement
uvFor uvx and for the Claude Code plugin, which starts its MCP server with uvx
The geo programThe skill needs it. The plugin does not install it
Claude CodeFor the skill and the plugin
An accountNone for audits. Only geo citations needs an API key from the AI provider you pick
  1. Install the geo program first. The pipx line below keeps it in its own environment. Without pipx, use the pip line further down.
  2. Check that it runs. The skill has Claude run the same check before anything else.
  3. Add the plugin in Claude Code. It brings the skill and the MCP server.
pipx install geo-optimizer-skill

The check from step 2 is geo --version, and the skill expects version 4.19.10 or later. Prefer pip, or want to try it without installing it, through uv’s uvx? These are the README’s lines:

pip install -U geo-optimizer-skill
uvx --from geo-optimizer-skill geo audit --url https://yoursite.com

Then run these two lines inside Claude Code. The README says one step is enough on Claude Code 2.1.275 or later, and setup/claude-code.md in the zip has that line:

/plugin marketplace add Auriti-Labs/geo-optimizer-skill
/plugin install geo-optimizer@geoready

How to use it

Ask Claude in plain words to audit a site for AI search. The skill’s rules decide the order, and some steps are yours.

Who does what
StepClaudeYou
PreflightRuns geo --version and stops if the program is missingInstall it
AuditRuns geo audit with --no-plugins and reads the scoreName the site
FixRuns geo fix as a preview and adds --apply only after you agreeSay yes, then upload the files
SchemaRuns geo schema to print JSON-LD, and adds --inject only after you agreeName the HTML file and agree
CitationsAsks before geo citations, which spends API creditSupply a key and approve

Fetched pages, robots.txt, llms.txt and sitemaps count as data. The skill tells Claude never to follow instructions found in them.

Example

You launched a small documentation site and want to know whether AI search can use it.

Audit https://yoursite.com for AI search visibility, then show me what you would fix

Claude runs geo audit and shows the score from 0 to 100 with a prioritized list of fixes.

It then previews the fix files and writes nothing until you agree. After a yes, geo fix --apply puts robots.txt rules, llms.txt, JSON-LD and meta tags into a geo-fixes folder for you to upload.

For newcomers and experts

If you are new. AI answer engines pick a few pages and quote them. GEO means making your site easy to reach and easy to quote. Two files matter early. robots.txt tells crawlers whether they may come in, and llms.txt is a short map for AI agents. The README calls llms.txt a signal that is not a proven ranking factor. Start with the audit and read the top fixes.

For experts. The default rubric v2 scores nine categories. The older rubric stays available for comparison with geo audit --score-version 1.

The nine scored categories
CategoryPointsWhat it looks at
Google AI readiness20Indexing, snippets, canonical, dates, byline, links, viewport and sitemap
Robots.txt14Googlebot and the citation bots
Schema JSON-LD14Validity, richness and a visible name match
Content14Headings, statistics, links, structure and image alt text
Brand and entity12Brand coherence, Knowledge Graph links, about and contact pages
Meta tags11Title, description and Open Graph
llms.txt6Found, sections, links and llms-full.txt
Signals6Page language, a feed and freshness
AI endpoint files3ai.txt, summary.json and markdown negotiation

Ten more checks, such as prompt injection and RAG chunk readiness, are informational. Only the negative signals can deduct points, up to 5. These lines are from the README:

geo audit --url https://yoursite.com --threshold 70
geo audit --sitemap https://yoursite.com/sitemap.xml --max-urls 25
geo llms --base-url https://yoursite.com --output ./public/llms.txt
geo citations --brand "Acme" --domain acme.com

Good to know

The program is free and MIT licensed, and nothing here needs the maker’s hosted service.

  • The skill is a rulebook. The program is the pip package, so install it before the skill can work. The plugin does not install it
  • Plugins: geo audit loads every installed check plugin unless Claude passes --no-plugins. Several other commands load them with no opt-out
  • Writing files: geo fix previews by default. With --apply it writes into ./geo-fixes/ and never edits your live site. geo schema --inject edits the HTML file you name, after saving a .bak copy
  • Citations: answers vary from run to run, so --runs 5 samples each question five times. Only some providers show their sources
  • Privacy: the skill says geo and geo-mcp have no telemetry and no update check. They fetch the address you audit and the AI provider you set up
  • Mismatches: docs/getting-started.md describes an older clone install. The README counts 12 MCP tools and docs/mcp-server.md says 10. The Astro integration’s own README says npm install where the main README says copy the folder
  • Maker: Juan Camilo Auriti at Auriti Labs. The names GEO Optimizer and GeoReady are covered by the maker’s name policy, and Mindket’s notes in the zip are not official

README and files

README
# GEO Optimizer: setup notes

This kit holds setup notes written by Mindket. It does not contain the geo program, the skill file, the plugin files, the MCP server or the context files. Those come from Auriti Labs’ repository and its Python package: https://github.com/Auriti-Labs/geo-optimizer-skill

License: MIT, Copyright (c) 2026 Juan Camilo Auriti. Full text: https://github.com/Auriti-Labs/geo-optimizer-skill/blob/main/LICENSE

The maker also publishes a name and logo policy (https://github.com/Auriti-Labs/geo-optimizer-skill/blob/main/TRADEMARKS.md). It says you may state that something is based on or compatible with GEO Optimizer, with a link to the repository, and asks that nobody present another product as GeoReady or as the official GEO Optimizer. These notes are Mindket’s and are not official.

The pages these notes came from: the README (https://github.com/Auriti-Labs/geo-optimizer-skill/blob/main/README.md), SKILL.md, the plugin files in https://github.com/Auriti-Labs/geo-optimizer-skill/tree/main/plugins/geo-optimizer and .claude-plugin, install.sh, capabilities.json, docs/mcp-server.md and docs/getting-started.md. Mindket copied every command from those pages and ran none of them. The few lines that are Mindket’s or a standard tool’s say so where they appear.

## What it is

A command-line program, geo, from the Python package geo-optimizer-skill. It scores a website from 0 to 100 on whether AI answer engines (ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews) can crawl it, read it and quote it, then lists what to fix. It can also write the fix files, check which AI engines mention a brand, and run as an MCP server.

The repository also holds a skill, geo-optimizer, which teaches Claude to run the program in the right order and with the right care. The skill does not contain the program. It has Claude check that geo is installed first, and stop if it is not.

## Files in this kit

- setup/command-line.md: running geo with uvx or pip, a first audit, and the fix preview
- setup/claude-code.md: the Claude Code plugin, the MCP server, and what the skill tells Claude to do
- setup/mcp-clients.md: the MCP server in Cursor, Windsurf and other clients
- setup/other-agents.md: the context files for Claude Projects, ChatGPT, Cursor, Windsurf and Kiro
- setup/ci-and-astro.md: the GitHub Action and the Astro integration

## What you need first

- Python 3.10 or newer, for the pip routes. It is the README’s stated requirement.
- pipx, only for the pipx line the SKILL.md offers beside pip. Without it, use pip.
- uv (https://docs.astral.sh/uv/) for uvx and for the Claude Code plugin, which starts its MCP server with uvx.
- The geo program itself, for the skill. The plugin does not install it.
- No account and no API key for audits. Only geo citations needs one, from the AI provider you pick.

## What installs where

- pip or pipx: the geo command, in that Python’s own environment. docs/mcp-server.md says the [mcp] extra adds the geo-mcp command and the MCP SDK.
- uvx: nothing stays installed. It runs the package from a temporary environment.
- The plugin: the skill file and an MCP server entry. Claude Code keeps the plugin files itself.
- The audit’s files: reports and fix files go to the folder you run it from. The SKILL.md names ~/.geo-optimizer/tracking.db and ~/.geo-cache as places the history and cache options write.

## Your data

The README says the program fetches only the address you audit and any AI provider you set up with your own key. The SKILL.md says geo and geo-mcp have no telemetry and no update check. Page addresses are checked against private network ranges before any fetch. geo citations sends the provider only the question texts, built from your topic or brand, and reads your key from the environment. What you type into your agent goes to that agent’s provider as in any other work.

## Things that do not match

- docs/getting-started.md describes an older install: it clones the repository to ~/geo-optimizer-skill and makes a ./geo wrapper and an update.sh. The README, SKILL.md and install.sh now install the Python package with pip. These notes follow the README.
- The README and SKILL.md count 12 MCP tools. docs/mcp-server.md’s heading says 10.
- The repository’s root .mcp.json is not the install route. It points at a local .venv/bin/geo-mcp and at a second tool, graft. The plugin has its own .mcp.json.

## Update and remove

The README gives no update or removal step. For the pip route, pip’s own commands apply: pip install -U geo-optimizer-skill is the README’s upgrade line, and pip uninstall geo-optimizer-skill removes the package. The second one is pip’s standard command, not the maker’s. The plugin updates and uninstalls through Claude Code’s /plugin menu.
Files
README.md  5 KB
setup/ci-and-astro.md  2 KB
setup/claude-code.md  4 KB
setup/command-line.md  3 KB
setup/mcp-clients.md  3 KB
setup/other-agents.md  3 KB

6 files, 19 KB, downloaded as a zip.

What’s new

  • Version 2 · 2026-11-09

    Updated by Mindket to the current version of this kit.

  • Version 1 · 2026-11-07

Reviews

No ratings yet

Write a review

No reviews yet.

What people say

Posts and comments that name this project on Hacker News, Bluesky, X, Reddit, Facebook and YouTube, found by a weekly search and on Trendshift. Mindket doesn’t write or check them. Open one to read it where it was posted.

juanauritiPre-registered a calibration study for my GEO tool. Result: null1 pointTom Dörrhttps://github.com/Auriti-Labs/geo-optimizer-skillTom Dörrhttps://github.com/Auriti-Labs/geo-optimizer-skill

Searched

More from Mindket

See all from Mindket

Nanosaur2-670M

well9472 · Kit

Makes illustrations from text prompts with a small 670M-parameter diffusion model, run through ComfyUI. Built as a low-budget research model for people who wan…

New

waqf-ocr-hand-written-v1

Waqf-AI · Kit

Reads handwritten Arabic text from images and returns it as text. A fine-tuned PaddleOCR-VL model for people digitizing Arabic handwriting.

New

Wan2.2-TI2V-5B-Diffusers

Wan-AI · Kit

Makes video clips from a text prompt or an image plus text, at up to 720P and 24 frames per second, in the Diffusers format for a single consumer GPU.

New

Similar projects

See all in Marketing & SEO

System Prompts Leaks

asgeirtj · Kit

A curated collection of verbatim system instructions for popular AI models and chatbot platforms, providing insight into how AI agents are configured to perfor…

New71k

Hermes Agent

NousResearch · Kit

Nous Research’s open source agent that you run in a terminal, in a desktop app or from a chat app. It keeps memory, writes skills from experience, runs schedul…

New261k

OpenHands Agent Canvas

OpenHands · Kit

OpenHands’ open source, self-hosted control center for coding agents. One browser app runs the OpenHands agent, Claude Code, Codex, Gemini CLI, Pi or OpenCode…

New92.8k