Guides for website readability and MCP monitoring.
Start with the website guides if you run a business site; the monitoring guides are for teams that run an MCP server.
All guides
- What AI assistants read on your website A practical 10-minute check for HVAC and home service websites: clear services, service areas, contact details, links and booking steps.
- AI website readability: 10 questions answered Straight answers for home service businesses about AI website scans, llms.txt, crawler controls, bookings and what a passing check means.
- Onbiom, DIY, an agency or AI search monitoring? Compare Onbiom, DIY, SEO agencies, OtterlyAI, Peec AI and Profound for a home-service website. Choose checks, hands-on help or AI visibility tracking.
- Which AI crawlers your robots.txt lets in What OAI-SearchBot, GPTBot, ChatGPT-User, Google-Extended, ClaudeBot, Claude-SearchBot and PerplexityBot do, what blocking each means, and a copy-paste robots.txt.
- Do you need llms.txt? What the evidence shows What llms.txt is, which vendors say they use it (and which say nothing), a minimal example file and what the Onbiom free check reports about it.
- Serving Markdown to AI agents: how it works and how to check it How a website can answer Accept: text/markdown with a Markdown page, what is and is not documented by vendors, one curl command to test it and what the Onbiom free check reports.
- Help AI agents find your MCP server: server cards and /.well-known discovery What an MCP server card and an AI catalog are, their documented status, a minimal example, what is known about client support and which discovery checks the Onbiom free scan reports.
- Monitor an authenticated MCP endpoint Configure endpoint access and separately observe MCP connection and tool inventory with a bearer token or supported preregistered OAuth flow.
- Understand MCP tool inventory changes Read added, removed or changed tool evidence, inventory response time and sampled availability without treating them as tool execution.
- Approve a fixed MCP tool check Preview fixed tool arguments, enable the exact request separately and distinguish output assertions, saved expectations and optional model advice.