QuantAI

GitHub SEO and GEO Pattern Notes

Last reviewed: 2026-07-04

These notes sit in the reference layer so the public entry pages can stay focused on users.

Review Method

This file is based on public online references, not private intuition. The review used GitHub repository search and direct inspection of README files, GitHub Pages output, llms.txt, robots rules, sitemap generation, and reference/spec layers.

Search patterns checked:

Selection criteria:

  1. The repository is public and has visible README or GitHub Pages content.
  2. The repository explicitly uses SEO, GEO, AEO, llms.txt, agent readiness, AI search, or GitHub Pages discovery.
  3. The project exposes enough structure to learn from: user-facing copy, machine-readable files, reference docs, schema, sitemaps, or well-known endpoints.

Repositories Reviewed

Source What it does Files or surfaces inspected Pattern worth using for QuantAI
Auriti-Labs/geo-optimizer-skill Open-source AEO/GEO toolkit with a product website and GitHub Pages docs. README.md, frontend/public/llms.txt, frontend/public/robots.txt, frontend/public/sitemap.xml, frontend/public/site.webmanifest, favicon and OG assets, sitemap generator. Keep the public README and site direct and benefit-led, then place bot rules, scoring logic, AI context, schema notes, and discovery files in a lower technical layer. Their sitemap script derives honest lastmod from source changes and exposes images in sitemap entries.
thedaviddias/llms-txt-hub Directory of websites and tools implementing the proposed llms.txt standard. README.md, apps/web/app/(files)/llms.txt/route.ts, scheduled index update workflow, About page metadata. A directory-style README uses brand name, one-sentence value, favicon, site link, and direct llms.txt / llms-full.txt links. The /llms.txt output is a plain-text generated index, not a keyword dump.
AnswerDotAI/llms-txt The original public proposal for /llms.txt. README.md, nbs/index.qmd, nbs/llms.txt, nbs/llms-sample.txt. Use the proposed format: H1, blockquote summary, short context, H2 file-list sections, and an Optional section for lower-priority resources. Use concise descriptions next to links.
sceneview/sceneview Open-source 3D/AR SDK that explicitly markets itself as AI-first while remaining user/developer-facing. README.md, root llms.txt, docs/docs/llms-full.txt, docs/docs/robots.txt, docs homepage, structured-data.json. The top-level README leads with product positioning, demo app downloads, quick examples, and platform matrix. AI assets stay in root/docs discovery files. This is the right shape for QuantAI: user app value first, agent-readable material deeper.
dodopayments/dualmark AEO infrastructure that pairs human HTML pages with markdown versions for AI agents. README.md, docs app routes, /llms.txt route, /raw/*.md routes, examples. Preserve a polished human page and expose clean markdown twins or raw markdown for agents. The useful principle is “same content, different rendering”, not aggressive AI-targeted rewriting.
jdevalk/specification.website Platform-agnostic website specification covering SEO, accessibility, security, well-known URIs, i18n, and agent readiness. README.md, public/.well-known/*, public/robots.txt, spec categories for SEO, i18n, well-known, and agent readiness. Treat agent readiness as a standards layer: stable URLs, structured data, language metadata, security contact, well-known files, and sources belong in predictable inner paths. Every claim should be sourced or marked as unsettled.

Online Verification Snapshot

Verified online on 2026-07-04 using GitHub repository pages and GitHub file-tree inspection. The purpose was to learn from real public implementations before shaping QuantAI copy and discovery files.

Repository Public signal checked Concrete files or paths found
Auriti-Labs/geo-optimizer-skill Repo description explicitly targets AEO, GEO, AI SEO, LLM SEO, citation tracking, ChatGPT, Perplexity, Gemini, and Google AI Overviews. Repository topics include geo, generative-engine-optimization, llms-txt, schema-markup, ai-seo, ai-visibility, and answer-engine-optimization. README.md, frontend/public/llms.txt, frontend/public/llms-full.txt, frontend/public/robots.txt, frontend/public/sitemap.xml, frontend/public/site.webmanifest, frontend/public/assets/og-image.png, references/schema-templates.md, site/.well-known/ai.txt.
thedaviddias/llms-txt-hub Repo description positions it as a directory for AI-ready documentation and tools implementing the proposed llms.txt standard. Topics include llms-txt, robots-txt, directory, nextjs, and cursor-ai. README.md, apps/web/app/(files)/llms.txt/route.ts, apps/web/app/robots.ts, apps/web/app/sitemap.ts, apps/web/app/projects/page.tsx, packages/content/data/websites/*-llms-txt.mdx, .github/workflows/update-llms-index.yml.
AnswerDotAI/llms-txt Original public proposal repository for /llms.txt; its homepage is llmstxt.org. README.md, nbs/index.qmd, nbs/llms.txt, nbs/llms-sample.txt, nbs/llms-ctx.txt, nbs/llms-ctx-full.txt, llms_txt/core.py, llms_txt/txt2html.py.
sceneview/sceneview Repo description is product-first while explicitly noting AI-first discovery assets: llms.txt, MCP server, and Copilot/Cursor rules. It also has app-store metadata and store badge assets. README.md, root llms.txt, docs/docs/llms-full.txt, docs/docs/robots.txt, docs/docs/structured-data.json, docs/docs/manifest.json, website-static/.well-known/llms.txt, website-static/llms-full.txt, website-static/robots.txt, website-static/sitemap.xml, website-static/assets/brand/stores/app-store.svg.
dodopayments/dualmark Repo description centers on AEO infrastructure, markdown twins for AI agents, and human HTML pages selected through content negotiation. Topics include aeo, ai-search, answer-engine-optimization, llms-txt, ai-bots, ai-seo, generative-engine-optimization, and geo. README.md, apps/docs/app/llms.txt/route.ts, apps/docs/app/raw/index.md/route.ts, apps/docs/app/raw/docs/[[...slug]]/route.ts, apps/docs/app/robots.ts, apps/docs/app/sitemap.ts, spec/content-negotiation.md, spec/llms-txt-extensions.md.
jdevalk/specification.website Repo description covers website SEO, accessibility, security, and agent readiness as standards, not growth-hack copy. Topics include seo, agent-readiness, llms-txt, mcp, well-known, and web-standards. README.md, public/robots.txt, public/site.webmanifest, public/.well-known/ai-catalog.json, public/.well-known/security.txt, public/.well-known/agent-card.json, src/pages/llms.txt.ts, src/pages/llms-full.txt.ts, src/pages/sitemap-index.xml.ts, src/content/spec/seo/*, src/content/spec/i18n/*, src/content/spec/agent-readiness/*.

Writing Patterns Learned

1. Keep the outer layer human and product-led

The strongest examples do not open with a raw SEO/GEO implementation diary. They open with the audience, category, value proposition, proof, and next action:

For QuantAI this means the top layer should say that QuantAI is an AI market research and investment analysis assistant for stocks, crypto, commodities, prediction markets, and portfolio research. It should not lead with “robots.txt, JSON-LD, llms.txt, sitemap” language.

2. Put technical SEO/GEO in an inner reference layer

The better repositories make the machine-readable and technical layer easy to find without forcing users to read it:

This matches the current QuantAI rule: user-facing pages stay readable; SEO/GEO mechanics live under docs/reference/ or machine-readable endpoints.

3. Use curated llms.txt, not keyword stuffing

The llms.txt examples are short indexes, not full landing pages. They make it easy for an LLM or agent to choose the right document:

For QuantAI, llms.txt should point to homepage, research page, iOS download page, pricing, languages, llms-full.txt, pricing.txt, schemas, and the AI catalog. It should not repeat every keyword variant.

4. Make multilingual and app-download signals real

GitHub Pages and README content can support multilingual discovery only if each locale has real metadata and useful copy:

For QuantAI, locale pages should describe the product in the language’s own search intent instead of simply translating a keyword list.

5. Separate claims from evidence

The useful examples cite research, standards, or documented methodology. The risky examples over-promise “rank in ChatGPT” or “guaranteed AI visibility.” For QuantAI:

External Guidance Checked

QuantAI Implementation Principles

  1. User-facing pages first: README, homepage, research, iOS, pricing, and language pages should explain QuantAI plainly.
  2. Reference layer second: keyword maps, schema JSON-LD, search submission notes, AI visibility notes, and this pattern file live under docs/reference/.
  3. Machine-readable access: keep llms.txt, llms-full.txt, pricing.txt, humans.txt, .well-known/security.txt, .well-known/ai-catalog.json, robots.txt, sitemap output, canonical URLs, JSON-LD, alternate language links, and App Store QR assets crawlable.
  4. No ranking promises: GitHub Pages, llms.txt, schema, and multilingual pages improve discoverability and extractability, but Google indexing and ranking are never guaranteed.
  5. No keyword stuffing: use natural product language at the top level; keep long keyword matrices in the reference layer.
  6. Use public domain contact only: support@aiquant.io.
  7. Keep financial scope precise: QuantAI is research and analysis software, not a broker, trading bot, order router, portfolio manager, personalized financial advisor, or guaranteed signal service.

Evidence-to-Action Map

Evidence QuantAI action
Google emphasizes helpful, people-first, organized content for generative AI search. Keep homepage, README, research, iOS, pricing, and locale pages user-facing and useful.
llms.txt proposal favors concise markdown indexes and optional deeper resources. Keep root llms.txt short; put exhaustive context in llms-full.txt and docs/reference/.
Auriti and SceneView expose AI crawler, sitemap, manifest, icon, and OG layers alongside product pages. Maintain crawlable robots, sitemap, schema, OG image, QR code, and app/download assets without turning the public page into a technical manual.
llms-txt-hub uses directory entries with direct llms.txt / llms-full.txt links. Make QuantAI’s machine-readable files easy to discover from sitemap, robots, README, and reference index.
Dualmark separates human HTML from markdown twins for agents. Prefer user-facing HTML/Markdown plus separate agent-readable text files instead of forcing one page to serve every audience.
specification.website sources every standards claim and marks unsettled areas carefully. Put methodology, source reviews, and limitations in reference pages; avoid unsourced SEO claims.

What Not To Copy Blindly