Auriti-Labs/geo-optimizer-skill
Audits and tracks whether ChatGPT, Perplexity, Gemini and Google AI Overviews cite your site. CLI, Python and MCP.
The build guide · Found by AI
The AI visibility stack
Updated · Repository snapshot: 9 September 2026
Your next customer may be asking AI who to use. Start by measuring the answer, then work on the parts you can change.
This collection groups open-source and source-available tools by the job they do. Choose the part you need; installing the whole list is not a requirement for being named.
Repository stars, last-push dates and licence labels were checked on GitHub on 9 September 2026. They describe that snapshot, not a performance or security test. Hosting, model-provider fees and licence conditions may apply. Read the repository’s licence before deploying it.
Deploy: needs a server. Run local: a CLI, desktop app or build step. Drop-in: a library for an existing site. Reference: a collection to read.
Stars show interest. Judge quality yourself. “Check repository licence” means GitHub did not return one recognised licence label.
1. Measure it
Query the engines with buyer questions and read who they name. Answer monitors and technical crawlers do different jobs: use the former to measure answers and the latter to inspect your site.
mverab/eGEOagents
Measures and optimises content for ChatGPT, Perplexity, Gemini and Claude. Ships on PyPI as egeo.
alexpospekhov/searchstack-aeo
One CLI that monitors visibility across Google AI Overviews, ChatGPT, Perplexity, Claude and Grok, and is happy on a cron.
promptclarity/promptclarity
A self-hosted Docker monitor of how AI mentions your brand across ChatGPT, Gemini and Perplexity, with the sources each engine cited.
StJudeWasHere/seonaut
A self-hosted technical audit crawler for finding crawlability and on-page SEO issues.
adityaarsharma/librecrawl-technical-seo-audit-mcp
A technical crawler exposed as an MCP server, so your own agent can check crawler access and structured data directly.
zubair-trabzada/geo-seo-claude
A GEO-first audit skill for Claude Code: citability scoring, AI crawler analysis, schema markup and platform-specific checks in one run.
amplifying-ai/awesome-generative-engine-optimization
A curated list of GEO resources if you want to go wider.
bmpi-dev/awesome-seo
A broader SEO research and traffic reading list for when you want the fundamentals underneath the AI layer.
2. Make your site AI-readable
Make your identity and content explicit. An llms.txt is a proposed format for an AI-readable site guide; Schema.org JSON-LD describes the entities and content on a page. Neither guarantees a citation.
AnswerDotAI/llms-txt
The canonical llms.txt spec from Answer.AI, plus a CLI that expands your file into a full context document for an LLM.
garmeeh/next-seo
Manages your meta tags and JSON-LD in Next.js, which are the signals an engine parses to work out what a page is.
google/schema-dts
TypeScript types for Schema.org JSON-LD. Check the resulting markup against the visible page and actual entity details.
4hse/astro-llms-txt
Generates llms.txt at build time for Astro sites, with no ongoing effort.
apify/actor-llmstxt-generator
Crawls a URL and produces an llms.txt for it, for people who do not want to run anything locally.
3. Be crawlable and renderable for agents
Some crawlers do not execute JavaScript. Check the HTML your server returns, then decide which crawlers you allow. Rendering and access controls are separate from whether an engine chooses to cite you.
stereobooster/react-snap
Pre-renders a single-page app to static HTML at build time, so a non-JS crawler sees the content.
prerender/prerender-node
Express middleware that serves a rendered HTML snapshot to crawlers. Its recent commits specifically add AI-crawler user-agents.
ai-robots-txt/ai.robots.txt
A maintained list of AI crawler user agents and blocking rules. Inspect it before use: copying the supplied rules can block the crawlers you wanted to allow.
unclecode/crawl4ai
A crawler and scraper for extracting pages as text for your own agent workflows. Its output is not a reproduction of every AI engine’s crawl.
firecrawl/firecrawl
A context API to search and scrape the web at scale. Useful for turning pages into the clean structured text agents ingest, and for watching how your content reads once stripped to text.
4. Publish citable content
Publish clear, sourced content in HTML that can be fetched. Static generators and content management systems give you different ways to maintain it.
gohugoio/hugo
A very fast static-site generator that outputs pure HTML with no JavaScript dependency.
withastro/astro
A content-first framework that ships near-zero-JavaScript HTML by default, with content collections built in.
jackyzha0/quartz
Turns Markdown or Obsidian notes into an interlinked static site.
TryGhost/Ghost
A publishing platform for articles and newsletters, with server-rendered pages and structured data.
payloadcms/payload
A headless CMS built on Next.js, so your content and your fast server-rendered pages live in one place.
strapi/strapi
A headless CMS, API-first, for rendering structured content into clean pages.
directus/directus
Turns a database into a content and data platform. Source-available under custom licence terms; check whether your use needs a paid licence.
5. Become agent-usable
MCP lets you expose tools and data to an AI client. Choose what it may read or change, and add authentication for private operations.
modelcontextprotocol/servers
The official reference collection of MCP servers you can copy as a starting point.
modelcontextprotocol/python-sdk
The official Python SDK for building MCP servers and clients.
modelcontextprotocol/typescript-sdk
The official TypeScript SDK, for teams already on Node or serverless.
jlowin/fastmcp
A fast, Pythonic framework for MCP servers with remote deploy and auth built in.
cloudflare/mcp-server-cloudflare
Cloudflare's own remote MCP servers on Workers, a production template for an endpoint agents can reach over the network.
github/github-mcp-server
GitHub's official MCP server, worth reading as a reference build for how a serious product exposes itself to agents.
aipotheosis-labs/aci
An open tool-calling platform that hooks 600+ tools into an agent through one unified MCP server, if you would rather connect than build.
punkpeye/awesome-mcp-servers
A large catalog of existing MCP servers. Check it before you build one, because someone may have already built it.
6. Monitor mentions and citations
Answer monitors show sampled mentions and citations. Analytics show visits, while server logs can show crawler requests. A crawler visit is not evidence of a recommendation.
allinurl/goaccess
A real-time log analyser, and a direct way to count GPTBot, ClaudeBot and PerplexityBot in your own server logs.
matomo-org/matomo
Self-hosted analytics whose log-import feature lets you build AI-bot segments, which the beacon-only tools cannot.
elmohq/elmo
A self-hostable platform that tracks how the engines mention and cite your brand, with every metric auditable in the code.
ai-search-guru/getcito
A self-hosted dashboard that tracks your visibility over time across ChatGPT, Perplexity, Claude and Google AI Overview.
letterstory/lettertrace
Open-source, bring-your-own-key answer-engine telemetry. You supply the model keys and it records how the engines answer for you over time, which keeps the running cost yours and the data in your hands.
7. Own your entities
Keep the names and identifiers for your organisation, people and products consistent. These general Wikidata and Schema.org tools help manage that information.
OpenRefine/OpenRefine
Cleans and reconciles organisation, people and product data against services such as Wikidata.
google/schema-dts
TypeScript types for Schema.org JSON-LD. Check the resulting markup against the visible page and actual entity details.
spencermountain/wtf_wikipedia
Parses Wikipedia markup so you can pull and watch your own entity data instead of finding out late that it drifted.
Being found is the goal.
Found by AI asks seven AI engines your buyers’ questions and keeps every answer word for word. Where an engine exposes its web searches, we keep those alongside the answer. Your AI can read your measurements over MCP.
The readiness method and audit suite are open source. The hosted visibility measurement is a service we run. Stay Found re-measures weekly and turns the results into a prioritised fix plan.
Start with the free scan: ChatGPT and Gemini, about a minute, no signup or card. Read the answers before you choose what to do next.
Read the readiness method and audit suite · How the measurement works
Scan your site free →