AI Visibility Check

What ai-visibility-check does

The ai-visibility-check MCP server evaluates whether a business website is cited when people ask AI answer engines for recommendations. It is intended for visibility checks based on customer-intent questions, such as requests for the best local provider in a particular category and location.

The check produces an overall score from 0 to 100 and assigns a visibility tier. It also preserves the detail behind that result: each query gets its own outcome, including whether the business site appeared, its citation position when present, and the sources cited instead.

How it works

The server sends each supplied question through a live Perplexity AI search. Results are therefore based on the answers and citations returned for those searches rather than on a static list of rankings.

Provide between one and five realistic questions that a potential customer might ask an AI system. Queries must not contain the business’s own name. This keeps the check focused on whether the business is discovered naturally for relevant demand instead of being prompted directly.

The returned data combines query-level findings with an aggregate visibility score and tier. A shareable branded scorecard link is also included for communicating the result.

Tools and capabilities

The ai-visibility-check MCP server exposes the check_ai_visibility tool. Its supported output includes:

• A visibility score from 0 to 100
• A visibility tier
• Results for each submitted question
• Whether the business website was cited
• Citation rank when the site was cited
• The competing sources selected by the AI answer engine
• A shareable branded scorecard URL

These results support both a high-level comparison and inspection of the individual questions that produced it. The competing-source information can show which other sites receive citations when the business is not included.

Limitations and notes

This check reflects the results of live Perplexity searches for the selected questions. It does not establish visibility for every possible customer query; the findings depend on the one to five questions supplied for the check.

Questions should represent genuine customer intent and should not name the business being evaluated. Including the business name would change the nature of the test and would not measure whether AI answer engines surface the business without that direct prompt.

The supplied material documents one MCP tool, check_ai_visibility, and does not specify installation commands, configuration variables, authentication requirements, transport details, or client-specific setup.