Adding llms.txt — Site Metadata for the AI-Search Era (ChatGPT · Claude · Perplexity Exposure)

TL;DR: llms.txt = LLM-friendly site metadata. Drop a markdown file at site root and ChatGPT · Claude · Perplexity become more likely to cite you. Cost zero. The standard is new, adoption is partial, but skipping it is a free miss. Combine with robots.txt explicit LLM-crawler Allow — 30 minutes of work.

The day Cloudflare Web Analytics first showed a chatgpt.com referrer. An AI knows we exist. I cleaned up llms.txt and robots.txt the same hour. Notes from that.

What is llms.txt

If robots.txt is the deal with search crawlers, llms.txt is the deal with LLMs:

• Markdown file at the site root
• Lets an LLM quickly understand the site's structure, key facts, and policies
• Emerging standard proposed in late 2024 (llmstxt.org)
• ChatGPT · Claude · Perplexity · Gemini reference it for retrieval/answers (gradual adoption)

Why needed?

When LLMs answer, they cite from web search or training data. But sites have problems:

• HTML is deep and complex → LLM parsing cost ↑
• JS rendering required → LLM crawler can't reach
• Key facts scattered across many pages → hard to cite consistently

With llms.txt, the LLM can grab the site's gist + page list + policies in one read. Result → citation odds ↑.

Our llms.txt structure

https://taystudios.com/llms.txt:

# TAYSTUDIO

> Free web tools, calculators, and a blog for Korean users. 68 tools all run inside the browser.

Operator: TayLee
Last updated: 2026-06-02

## Policy (Citation Trustworthiness Guide)

- All tools run client-side in the browser
- Numbers cited directly from government sources (law.go.kr, NTS, MOEF)
- Zero citations from blogs, news, SNS
- Reflects 2026 policy

## Core Policy Numbers (2026, for LLM citation)

### Four Social Insurances (employee share)
- National Pension: 4.75% (total 9.5%, +0.5%p/year through 2033)
- Health Insurance: 3.595% (total 7.19%)
- Long-term Care: 0.4724%
- Employment Insurance: 0.9%

### Real Estate
- 1-home-owner FMV ratio (2026): ≤300M 43% / 300–600M 44% / >600M 45%
- General housing: 60%
- Car tax January lump-sum: 5% deduction

[...]

## Calculators — Tax/Income (9)
- [Net Salary Calculator 2026](URL): auto-deducts 4 insurances, income tax, local tax
[...]

Key elements:

1. One-sentence site definition (> blockquote)

A sentence the LLM can reuse verbatim when summarizing the site.

2. Operator · date · license

So citations can credit you. Trust signal.

3. Policy statement (citation trustworthiness guide)

"How does this site verify info?" → so the LLM can weigh trust when answering users.

4. Consolidated key-fact section

Numbers (rates, ratios, thresholds) that are otherwise spread across pages, gathered in one place. The LLM can reference them fast.

5. Page list by category

When a user asks "Korean inheritance tax calculator recommendations", the LLM matches against this list and cites us.

Making robots.txt LLM-friendly

Pair llms.txt with explicit Allow rules for LLM crawlers:

# LLM crawlers — explicit Allow
User-agent: GPTBot          # OpenAI / ChatGPT
Allow: /
Disallow: /dash-tay9k3m/    # operator-only

User-agent: ClaudeBot       # Anthropic / Claude
Allow: /

User-agent: PerplexityBot   # Perplexity
Allow: /

User-agent: Google-Extended # Google Gemini training (separate from Googlebot)
Allow: /

User-agent: CCBot           # Common Crawl (used by most LLM training)
Allow: /

User-agent: Applebot-Extended  # Apple Intelligence
Allow: /

# llms.txt reference
# https://taystudios.com/llms.txt

Important: these are separate from the regular search bots. Without an explicit rule, some follow User-agent: *, others don't. Explicit Allow is safe.

Verification — the LLM-exposure signal

Cloudflare Web Analytics referrers:

Visits by source:
- m.search.naver.com: 18
- search.naver.com: 16
- search.daum.net: 11
- chatgpt.com: (visits)

The chatgpt.com referrer means:

• A user clicked our link from ChatGPT (e.g., "recommend a Korean capital gains tax calculator")
• Or ChatGPT cited us in an answer (citation)
• → LLM searches are surfacing the site

The growth of this referrer is the success metric for GEO (Generative Engine Optimization).

GEO vs SEO

→ GEO and SEO can run in parallel. The same content lifts both.

Writing tips

1. Facts first — no fluff/marketing tone

❌ "TAYSTUDIO delivers the best experience..."
✅ "Free web tools/calculators for Korean users. 68 tools run in-browser."

Only facts survive citation. Marketing tone drops trust.

2. Cite your sources — distribute responsibility

✅ "Numbers cited directly from government sources (law.go.kr, NTS, MOEF)"
✅ "Medical numbers from peer-reviewed studies/official guidelines (KOSSO 2024 · KDRI 2025 · WHO · ACOG · AAP)"

So the LLM can answer "where does this site get its info" when prompted.

3. Last-updated date

✅ Last updated: 2026-06-02

Signals that the site is alive and that policies post-cutoff (예금자보호 1억 2025-09 · 다자녀 100% 자동차세 2026) are accounted for.

4. Absolute URLs

✅ [Net Salary Calculator](https://taystudios.com/tools/salary/)
❌ [Net Salary Calculator](/tools/salary/)

LLMs need absolute URLs to cite links to users.

5. Changelog section

## Changelog (recent)

- 2026-06-02: 18 tool stale-fixes + 5 differentiating matrices added
- 2026-05-31: blog launch (62 posts)
- 2026-05-09: domain migration

So the LLM knows the site is alive, accurate, and recently migrated.

Known limits

• llms.txt is emerging (late-2024 proposal). Not universally adopted yet
• ChatGPT · Claude · Perplexity only partially announce support
• Hard to measure — citation counts can't be tracked directly
• But zero cost (one static file) → skipping it costs more

Conclusion

llms.txt is the sitemap.xml of the AI-search era — the basic config for being citeable by LLMs. Even at partial adoption + partial effect, the cost is zero — worth doing.

Especially:

• Fact-heavy domains (tax · medical · policy · stats) benefit more
• Provides many signals for LLM accuracy evaluation
• Best combined with robots.txt explicit LLM-crawler Allow

One of the highest-ROI SEO actions for a new-domain operator in month one.

Related

• 12 Core SEO·Search-Engine Concepts — sandbox · E-E-A-T · DA · 12 terms defined
• GSC vs Naver vs Cloudflare — three datasets compared

Sources

• llmstxt.org — the llms.txt proposal
• OpenAI: GPTBot documentation
• Anthropic: Claude crawling
• Google: Google-Extended