Post-deploy checklist for agentic-kit.dev crawl governance, CDN settings, and generative visibility monitoring.
| Check | Action |
|---|---|
| Cloudflare (if proxying Vercel) | Disable blanket “Block AI Scrapers”. Use granular rules so live-retrieval user-agents (OAI-SearchBot, ChatGPT-User, Claude-SearchBot, PerplexityBot) receive 200 responses. |
| Vercel Firewall | Confirm no rule returns 403 for allowed bots on /, /docs, /llms.txt. |
curl -sI -A "OAI-SearchBot" https://agentic-kit.dev/llms.txt
curl -sI -A "PerplexityBot" https://agentic-kit.dev/docs
curl -sI https://agentic-kit.dev/ | grep -iE '^(link|x-llms)'Expect 200 for allowed agents. Pre-training agents (GPTBot, ClaudeBot, CCBot, Google-Extended) are discouraged via public/robots.txt (Disallow: /).
Confirm discovery files are plain text / JSON, not SPA HTML:
curl -sI https://agentic-kit.dev/llms.txt | grep -i content-type
curl -s https://agentic-kit.dev/llms.txt | head -n 5
curl -sI https://agentic-kit.dev/.well-known/mcp/server-card.json | grep -i content-type- Submit
public/sitemap.xmlin Google Search Console and Bing Webmaster Tools. - Optional: IndexNow ping after deploy when URLs change.
Before shipping meaningful content changes:
- Google Rich Results Test :
/,/docs, one doc URL. - Schema.org Validator : paste JSON-LD from view-source.
- Local:
pnpm validate:jsonld
Most generative answers do not send click-through traffic. Use a citation-monitoring product for brand mention share, for example Otterly.AI, Peec AI, or Profound.
Suggested query themes:
- eve agent trust / authority diff
- Aletheia authority diff / honesty contract
- Kit Certified eve agent
- Agentic Kit policy packs
- inspect stamp ship eve agents
If on-site analytics are added later, filter referrers for:
chatgpt.com,chat.openai.comperplexity.aiclaude.ai,anthropic.comgemini.google.com
Treat spikes as directional only; they under-count answer-pane citations.
pnpm generate:llms
pnpm generate:discoverability
pnpm check:syncCommit updated public/llms*, public/robots.txt, public/sitemap.xml, public/ai.txt, and public/.well-known/.