X thread — Agent Readiness Report: Vercel
7 tweets. Tweet 1 leads with score + dominant theme + tags. NO link in tweet 1 (X throttles link-leading threads). Link lands in tweet 7. Tags @vercel and @rauchg on tweet 1.
Tweet 1 (266 chars)
@vercel scored 75/100. Level 3 — Agents First.
First product in the Agent Readiness Reports series to hit Level 3 on any surface.
But the AI SDK home — sdk.vercel.ai — scored 25/100, Level 1.
The variance is the story: 75 / 55 / 25.
@rauchg
Tweet 2 (270 chars)
vercel.com/docs is the high water mark.
Real /llms.txt at the docs root. Markdown content negotiation passes. OpenAPI surface discoverable.
Homepage hero references both MCP and the SDK alongside human onboarding.
10/10 on visibility. Almost no one passes that one.
Tweet 3 (273 chars)
Then sdk.vercel.ai — Vercel’s flagship AI product surface.
No Content-Signal. No markdown negotiation. No /AGENTS.md. No MCP card.
Triggers the Agents Without Rules anti-pattern.
The robots.txt itself says “Move to ai-sdk.dev.” The migration left the agent story behind.
Tweet 4 (264 chars)
Vercel’s robots.txt declares Content-Signal: search=yes, ai-input=yes, ai-train=no
Cloudflare invented the directive. Vercel uses it with the opposite default on training.
Both legitimate policy choices. The rubric credits either direction. The signal is what matters.
Tweet 5 (273 chars)
Fix #1: ship an MCP Server Card from vercel.com. Reference it from the homepage hero.
Vercel runs MCP servers in production through the platform.
An agent reading vercel.com sees “SDK” but cannot find the install command, the auth flow, the discovery file.
Worth 15 points.
Tweet 6 (270 chars)
Fix #2: lift the docs playbook to sdk.vercel.ai (and its destination ai-sdk.dev).
The docs surface already runs the pattern. Same hosting. Same team.
Apply Content-Signal, markdown negotiation, /AGENTS.md, MCP card to the AI SDK home.
Lowest-effort 50-point swing.
Tweet 7 (240 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/vercel/
Tweet count: 7. All tweets under 280 chars.
Tags: @vercel and @rauchg on tweet 1; @leeerob is a strong alt for tweet 1 if Rauchg has been mentioned recently. 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.