[0ptaq0]

AI agents (ChatGPT, Claude, Perplexity, coding agents like Claude Code or Cursor) read markdown much better than rendered HTML. vitepress-plugin-llms already builds a markdown copy of every docs page (/page.md); this makes agents find it and get it without guessing.

Changes:

• middleware.js (Vercel Routing Middleware): a page requested with Accept: text/markdown answers with its .md copy, the home page with llms.txt. Pages without a copy fall back to HTML. .md responses get a Link: rel="canonical" header pointing at the HTML page. Adds @vercel/functions (dev dependency, only used by the middleware).
• VitePress config: every page links its markdown copy with <link rel="alternate" type="text/markdown"> (home page: llms.txt).
• robots.txt: names the OpenAI, Anthropic and Perplexity crawlers next to * (same rules, everything allowed), adds Content-Signal: search=yes, ai-input=yes, ai-train=yes and points at llms.txt.

Verified with npm run docs:build: every rel="alternate" link points at a file in the build.

Try it on the preview or after deploy:

curl -H "Accept: text/markdown" https://trigger.beyondthecloud.dev/introduction
curl -I https://trigger.beyondthecloud.dev/introduction.md

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
trigger-lib	Ready	Preview	Oct 25, 2026 1:11pm UTC

[github-actions]

🚀 Salesforce CI

✅ Deployment succeeded

✅ All tests passed

==========================================
     APEX TEST EXECUTION SUMMARY
==========================================

📊 Total Tests: 1305
✅ Passed: 1305
❌ Failed: 0
⏭️  Skipped: 0


🎉 All tests passed successfully!

📦 Workflow run, logs and artifacts

---

📊 Stats: 1305 total | ✅ 1305 passed | ❌ 0 failed
🤖 Automated comment by Salesforce CI