HomeBlogAgent Protocol Tracker: Live Status of Every Agent-Readiness Standard
    Agent Protocol Tracker: Live Status of Every Agent-Readiness Standard
    BlogAugust 27, 20264 min read

    Agent Protocol Tracker: Live Status of Every Agent-Readiness Standard

    Share this article
    Share

    The living reference: specification stage, governance, real consumers, and top-500 adoption numbers for all 16 agent-readiness protocols. Updated monthly and after major protocol news.

    This page tracks every protocol in the agent readiness stack: its specification stage, who consumes it today, real adoption numbers from our top-500 sweep, and our current adopt/watch/skip verdict. We update it monthly and after major protocol news.

    Last verified: July 24, 2026. llms.txt row updated August 19, 2026 for the v2 spec. Next scheduled update: mid-September 2026.

    Specification status

    ProtocolStageGovernanceWho consumes it todayschema.orgMatureCommunityEvery AI systemrobots.txt AI rulesDe factoConventionAll major AI crawlersllms.txtv2 (August 2026)CommunityCoding agents, docsMarkdown negotiationEmerging conventionNoneClaude Code, OpenCode

    Interaction, commerce, and trust protocols move faster; check the changelog for what changed.

    ProtocolStageGovernanceWho consumes it todayMCP + server cardsDe facto standardAnthropic, openAll MCP clientsA2A agent cardsv1.0 (Apr 2026)Linux FoundationCopilot Studio, Azure AI Foundry, BedrockAgent SkillsOpen standard (Dec 2025)agentskills.io~40 clientsWebMCPW3C CG draft, Chrome 149 origin trialW3C WebML CGGemini in ChromeNLWebOpen project, adoption stalledMicrosoftVery few

    Money and trust close out the stack.

    ProtocolStageGovernanceWho consumes it todayUCPLive (Jan 2026)Google, Shopify coalitionAI Mode, Gemini, YouTubeACPLive (Sep 2025)OpenAI, StripeChatGPTx402Live, Foundation under Linux Foundationx402 FoundationAgent API buyersAP2v0.2, moving to FIDO AllianceGoogleGoogle surfaces onlyWeb Bot AuthIETF draft, WG charteredIETFClaude, ChatGPT, Perplexity sign; CDNs verifyRSLLive (Sep 2025)RSL coalitionLicensing crawlersPay Per UseLive (Jul 2026)CloudflareCloudflare-fronted sites

    Adoption among the web's top 500 domains

    From GenReady's sweep of the Tranco top 500 (328 reachable sites), July 12, 2026. Full analysis in the State of Agent Readiness report.

    SignalSitesSharerobots.txt AI crawler rules8124.7%llms.txt4614.0%Markdown negotiation92.7%MCP server card41.2%A2A agent card10.3%UCP manifest10.3%Agent Skills index00.0%x402 manifest00.0%

    Horizontal bar chart of agent-readiness adoption across the top 500 domains: robots AI rules 24.7%, llms.txt 14%, markdown negotiation 2.7%, MCP cards 1.2%, UCP and A2A cards under half a percent

    The ecosystem numbers look different from the top-500 numbers. UCP counts 8,000+ verified stores, roughly 99% on Shopify. x402 processed 169M payments across 100,000 sellers in year one. A2A runs in 150+ production organizations. Agent Skills counts ~40 compatible clients, growing by ~200 new skills weekly. Adoption is happening through platforms and APIs first, not through the web's biggest consumer domains.

    Current verdicts

    Act now on four things. schema.org, robots.txt AI rules, and markdown negotiation apply to every site. UCP applies if you sell online.

    Worth adding, case by case. llms.txt reached v2 in August 2026, which added discovery link relations and told publishers to point its links at markdown versions of their pages. It still shows no citation lift in AI search, but it earns its ten minutes if coding agents read your docs. Agent Skills and MCP server cards pay off if you have a product or API. RSL fits content with licensing value.

    Watch the rest. WebMCP, NLWeb, x402, AP2, Web Bot Auth, and Pay Per Use are not ready for a typical site's to-do list yet.

    Per-site-type guidance lives in the adopt/watch/skip matrix. The fastest path through the act-now list is the afternoon checklist.

    What we're watching next

    Three dates matter this quarter. Microsoft Build 2026 decides whether NLWeb gets distribution or stays stalled. August 2026 is the IETF's target for the Web Bot Auth operational document. On September 15, 2026, Cloudflare's new AI-blocking defaults take effect for new customers. That change should push more AI companies toward paid access rails.

    We also watch for four smaller signals. WebMCP graduating its origin trial, or a second browser shipping it. UCP expanding internationally. ACP reaching surfaces beyond ChatGPT. Any major crawler adopting Accept: text/markdown.

    Changelog

    August 19, 2026 — llms.txt moved from "de facto proposal" to v2. The August 2026 revision adds discovery link relations (rel="describedby" for the file, rel="alternate" type="text/markdown" for page markdown), allows page.md alongside page.html.md, defines subpath coverage with most-specific-wins, and drops the Optional section's mechanical meaning along with the llms_txt2ctx tooling. Consumer column changed to coding agents and docs, where the evidence actually sits.

    July 12, 2026 — Page created. First top-500 sweep run. Initial statuses recorded: WebMCP in Chrome 149 origin trial, A2A at v1.0, x402 Foundation under Linux Foundation, UCP at 8,000+ stores, Pay Per Use launched July 1.


    Check your own site against every signal on this page: run a free GenReady scan — under 60 seconds.

    Found this useful?

    Share it with someone who's trying to improve their AI visibility.

    Written by

    GenReady Team

    We help website owners understand how AI crawlers see their content - and how to improve it. Follow us for practical AI readiness tips.

    genready.ai →