Skip to the content.

X thread — Agent Readiness Report: Anthropic

7 tweets. Tweet 1 leads with score + dominant theme + tag. NO link in tweet 1 (X throttles link-leading threads). Link lands in tweet 7. Tag @AnthropicAI on tweet 1.


Tweet 1 (262 chars)

@AnthropicAI invented MCP — the protocol the Agent Readiness rubric is built around.

We pointed the scorer at Anthropic next.

60/100 on docs. Level 2.

Marketing site: 5/100. Level 0.

The MCP authors do not name a single Anthropic bot in their own robots.txt.


Tweet 2 (261 chars)

docs.anthropic.com is the high water mark.

Real /llms.txt — 146KB, structured per product, 11 languages.

Real /.well-known/mcp-server-card with OAuth-PKCE discovery.

20/20 on content-accessibility. Markdown content negotiation works.

60/100 earned honestly.


Tweet 3 (266 chars)

www.anthropic.com scores 5/100.

Generic robots.txt. No /llms.txt. No /AGENTS.md. No MCP card. No CLI install in the hero.

The marketing site for the company that defined the protocol the rubric credits is invisible to that protocol.

Same building. Different team.


Tweet 4 (265 chars)

The bots Anthropic blocks in claude.ai’s robots.txt: GPTBot, OAI-SearchBot, ChatGPT-User. All competitor.

The bots Anthropic operates — ClaudeBot, anthropic-ai, claude-web — appear in zero of the three robots.txt files.

The protocol authors don’t name themselves.


Tweet 5 (251 chars)

Fix #1: lift the docs pattern to anthropic.com.

/llms.txt. /AGENTS.md. MCP server card. Sitemap.

Same files, one subdomain over. Worth ~50 points on the marketing root.

Closes the Invisible Product anti-pattern in one PR. Same fix Cloudflare needs.


Tweet 6 (260 chars)

Fix #2: ship a robots.txt that names ClaudeBot, anthropic-ai, claude-web.

Allow or Disallow — the value is the explicit declaration.

Plus Content-Signal: ai-train=yes, search=yes, ai-input=yes.

15 points across every surface. The protocol should be legible.


Tweet 7 (187 chars)

Second in a bi-weekly Agent Readiness Reports series. Scoring named products against the Agents First framework.

Full report — rubric, raw probe data, top fixes:

https://agentsfirst.dev/reports/anthropic/


Tweet count: 7. All tweets under 280 chars.

Tags: @AnthropicAI on tweet 1; report URL only on tweet 7 (X throttles link-leading threads ~40%).

Note: tweet body is the text shown to the reader between > lines, including blank lines as \n chars. The > markdown markers and “Tweet N” labels are NOT part of the tweet.