Faro AI Signals·July 7, 2026

Anthropic Expands MCP Registry as llms.txt Crosses 50K Deployments

MCP's first official third-party registry launched this month, making agent-facing service discovery a live infrastructure problem rather than a theory. Meanwhile, llms.txt adoption hit a measurable milestone, signaling that crawlers are now actively differentiating between sites that have it and those that don't.

ShareShare

This Week's Signals

1

Anthropic Opens MCP Registry to Third-Party Servers, Enabling Agent-Driven Business Discovery

Source: Anthropic

Why it matters for your score

The official MCP registry means Claude and compatible agents can now discover, authenticate, and invoke business-exposed tools without custom integrations — if your service is registered. Businesses without an MCP-compatible endpoint are invisible to this entire layer of agent-driven commerce. This is the moment 'agent-facing API' goes from a buzzword to a checkbox that determines whether you appear in an agentic workflow at all.

2

llms.txt Deployments Surpass 50,000 Verified Domains, Crawlers Show Preferential Indexing Behavior

Source: llmstxt.org

Why it matters for your score

Independent tracking of llms.txt adoption shows the file is now present on over 50,000 domains, with early crawler log analysis from Perplexity and ClaudeBot indicating these bots are spending more time on structured content blocks declared in llms.txt than on equivalent unstructured pages. If your business lacks this file, crawlers are likely deprioritizing your content in favor of competitors who have declared it cleanly. The gap between adopters and non-adopters is now measurable in citation frequency.

3

Perplexity Confirms 'Verified Business Answers' Program Weights Structured Schema and Pricing Transparency

Source: Perplexity AI

Why it matters for your score

Perplexity's business answer cards — the structured recommendation boxes that appear above organic citations — now explicitly factor in the presence of JSON-LD schema, published pricing data, and a valid llms.txt when ranking which businesses to surface for commercial queries. This is the clearest public confirmation to date that AI answer engines are using structured, machine-readable signals as ranking inputs — not just crawling prose. Businesses with incomplete schema or hidden pricing are measurably less likely to appear in these high-intent placements.

What to do this week

  1. 1

    Register your service with Anthropic's MCP registry this week — even a minimal endpoint listing your core offering, hours, pricing, and booking URL makes you discoverable to Claude-based agents running commerce and scheduling tasks. Use Faro's /tools/ai-readiness-scan to identify whether your current site structure is MCP-ready before submitting.

  2. 2

    Generate or update your llms.txt file using Faro's /tools/llms-txt generator, then verify it's accessible at yourdomain.com/llms.txt — with crawler log evidence now showing preferential indexing, having a malformed or outdated file is actively worse than a well-structured one. Include a clear service description, pricing tier summary, and a link to your most authoritative page.

  3. 3

    Audit your JSON-LD schema for pricing completeness using Faro's /tools/ai-schema tool — Perplexity's confirmed weighting of pricing transparency means any business with 'contact us for pricing' or missing PriceRange markup is disqualifying itself from Verified Business Answer placements. Add a LocalBusiness or Product schema block with explicit priceRange or offers fields before next Monday's crawl cycle.

Sources

← All editionsScan your site now →