Anthropic's MCP Goes GA: Every Business Now Has an Agent Surface
Anthropic formally released Model Context Protocol as a stable 1.0 specification, moving MCP from experimental to production-ready infrastructure. Combined with accelerating llms.txt indexer adoption and new structured-citation behavior in Perplexity's Agent Mode, the gap between AI-visible and AI-invisible businesses just got measurably wider.
This Week's Signals
Anthropic Ratifies MCP 1.0 Spec, Enabling Stable Agent-Facing Business Endpoints
Why it matters for your score
With MCP 1.0 stable, Claude and third-party agents can now reliably connect to business endpoints — inventory, pricing, booking, services — without bespoke integrations. Businesses that publish a discoverable MCP server become callable resources for any compliant agent, not just Anthropic's. If your business has no MCP surface, agents default to competitors who do.
Perplexity Agent Mode Now Cites Businesses With Structured Pricing Data Over Plain-Text Pages
Why it matters for your score
Perplexity's updated Agent Mode retrieval stack demonstrably favors pages where pricing, services, and hours are marked up with JSON-LD schema over equivalent pages with the same information in unstructured prose. For local and e-commerce businesses, this is the same forcing function that Google's rich results created in 2015 — except the citation either happens or it doesn't, with no middle ground of 'ranked lower.' Run Faro's /tools/ai-schema audit to confirm your PriceSpecification and LocalBusiness schema is well-formed and machine-readable.
llms.txt Crawler Indexing Reaches Critical Mass as Twelve Major Platforms Confirm Active Parsing
Why it matters for your score
The llms.txt registry now confirms active parsing by twelve distinct AI platforms and toolchains, including retrieval pipelines used by enterprise copilots built on GPT-4o and Gemini 1.5. A well-formed llms.txt file is no longer speculative infrastructure — it is a direct input to the context windows agents use when deciding which business to recommend. Businesses without one are absent from a growing share of agentic lookups, regardless of how strong their traditional SEO is.
What to do this week
- 1
Publish or update your llms.txt file this week to reflect current services, pricing tiers, service area, and a plain-language business summary — twelve platforms are actively parsing these files, and stale or missing entries are now a confirmed ranking liability. Use Faro's /tools/llms-txt generator to produce a spec-compliant file in under five minutes, then verify it resolves cleanly at yourdomain.com/llms.txt.
- 2
Audit your JSON-LD schema for PriceSpecification, LocalBusiness, and Service markup using Faro's /tools/ai-schema tool — Perplexity Agent Mode's confirmed preference for structured pricing data means unstructured price pages are being skipped in favor of competitors who markup correctly. Fix any missing 'priceRange,' 'areaServed,' or 'openingHoursSpecification' fields before next Monday.
- 3
Assess your MCP readiness now that the 1.0 spec is stable: run Faro's /tools/ai-readiness-scan to identify whether your site exposes any agent-callable endpoints and whether your robots.txt permits MCP-adjacent crawlers including ClaudeBot and PerplexityBot. If your scan shows no MCP surface and crawler blocks, you have a concrete two-step remediation path to address before the next wave of agentic traffic indexes your category.