Installation
Installation connects your existing URLs to camolabs. Your existing site stays the default response. Only eligible agent requests are routed to camolabs for an Agent Page.
Installation has two pieces, both set up in your CDN or hosting layer:
- An Agent domain, such as
agents.yourdomain.com, pointed at camolabs with a CNAME toapp.camolabs.ai. - A routing rule on your main domain. It passes normal traffic to your existing site and routes only eligible agent requests to the Agent domain.
Human browsers and search crawlers always pass through to your existing site.
Set it up
The step-by-step lives in the guide for wherever your public site is served:
For any other CDN, follow the same model: pass traffic to the existing site by default, route only eligible GET page requests to the Agent domain, and return the camolabs response only when it is 200 text/html with x-camolabs-agent-page-outcome: served_agent_page.
Values you will use
| Value | Example | Used in |
|---|---|---|
| Main domain | www.yourdomain.com | Existing public site |
| Agent domain | agents.yourdomain.com | DNS and Settings > Install |
| CNAME target | app.camolabs.ai | DNS provider |
CAMOLABS_AGENT_ORIGIN | https://agents.yourdomain.com | Cloudflare Worker or Vercel Middleware |
| Routing policy | https://agents.yourdomain.com/.well-known/camolabs-routing-policy.json | Edge adapter |
| MCP endpoint | https://agents.yourdomain.com/mcp | Approved Actions |
What gets routed
camolabs never blocks anything. The default path is pass-through to your existing site. Requests that look automated can receive Agent Pages when they hit a published route.
These AI agents receive Agent Pages when they hit a published route:
| Agent | User-agent token |
|---|---|
| ChatGPT | ChatGPT-User |
| OpenAI Search | OAI-SearchBot |
| Claude | Claude-User |
| Claude Search | Claude-SearchBot |
| Perplexity | Perplexity-User |
| PerplexityBot | PerplexityBot |
| Google Agent | Google-Agent |
| Meta Fetcher | Meta-ExternalFetcher |
| Mistral | MistralAI-User |
| Mistral Index | MistralAI-Index |
| Amazon User | Amzn-User |
| Amazon Search | Amzn-SearchBot |
These AI crawlers are labeled in your activity but still get your existing site, not Agent Pages:
| Crawler | User-agent token(s) | What it is |
|---|---|---|
| GPTBot | GPTBot | OpenAI training |
| ClaudeBot | ClaudeBot, anthropic-ai | Anthropic training |
| GoogleOther | GoogleOther | Generic Google |
| Meta AI | Meta-ExternalAgent | Meta crawler |
| Amazonbot | Amazonbot | Amazon crawler |
Search crawlers always continue to your original site to avoid cloaking:
| Search crawler | User-agent token(s) |
|---|---|
| Googlebot | Googlebot |
| Bingbot | bingbot, BingPreview |
| Yahoo Slurp | Slurp |
| DuckDuckBot | DuckDuckBot |
| Baiduspider | Baiduspider |