Put your agent files where scanners look

# .github/workflows/agentready.yml
name: AgentReady audit
on:
  pull_request:
  push:
    branches: [main]
  schedule:
    - cron: "0 6 * * 1" # every Monday

jobs:
  agent-readiness:
    runs-on: ubuntu-latest
    env:
      SITE_URL: https://your-site.com # change me
    steps:
      - name: llms.txt is served as plain text
        run: |
          code=$(curl -s -o llms.txt -w "%{http_code}" "$SITE_URL/llms.txt")
          test "$code" = "200" || { echo "llms.txt returned $code"; exit 1; }
          if grep -qi "<html" llms.txt; then echo "llms.txt is an HTML page, not a file"; exit 1; fi
      - name: agent.json is valid JSON
        run: curl -sf "$SITE_URL/.well-known/agent.json" -o agent.json || curl -sf "$SITE_URL/agent.json" -o agent.json && jq . agent.json > /dev/null
      - name: robots.txt does not block AI crawlers
        run: |
          curl -sf "$SITE_URL/robots.txt" -o robots.txt || exit 0
          for bot in GPTBot ClaudeBot PerplexityBot Google-Extended; do
            if awk -v b="$bot" 'tolower($0) ~ "user-agent: *"tolower(b){f=1;next} /user-agent/{f=0} f && /^disallow: *\/ *$/{exit 1}' robots.txt; then :; else echo "$bot is blocked"; exit 1; fi
          done
      - name: AI crawlers are not challenged by the firewall
        run: |
          code=$(curl -s -o /dev/null -w "%{http_code}" -A "GPTBot/1.0" "$SITE_URL/")
          test "$code" -lt 400 || { echo "GPTBot gets HTTP $code"; exit 1; }