Vantage

What the Vantage MCP server does

The Vantage MCP server examines how AI-generated search answers cite websites and brands. It can check whether a domain appears for selected prompts, identify the domains receiving the most citations for a topic, and track a domain’s monthly mention counts. The supplied description names ChatGPT, Perplexity, and Gemini as supported AI systems; the README also references Google AI Overview.

Results focus on observed citations rather than general SEO metrics. The server separates a brand being named from a page being linked, which helps distinguish recognition from an actual citation. It also reports answer-shape details, such as whether the response begins with a list, how long its opening is, and how many sources it cites.

How it works

The hosted service is exposed at https://vantagemcp.dev/mcp and uses a bearer API key in the MCP client configuration. Calls consume quota according to the operation. Usage lookup costs no units, prompt coverage costs one unit per keyword, and the deprecated basic visibility check costs ten units. Several analysis operations can process multiple topics in one request: prompt coverage supports up to 10 keywords, and batch citation-structure analysis supports up to 10 related topics.

The Vantage MCP server can also run locally. Its Python module starts an MCP process over stdio by default, or an HTTP service when launched with --http. Local operation still requires a DataForSEO account because that service supplies the underlying SERP and AI-answer data. Billing and account provisioning for Vantage itself are not included in the repository.

Setup and configuration

For hosted use, add the MCP endpoint and an Authorization: Bearer ... header to the client configuration, then obtain a free API key from Vantage. The stated free allowance is 30 quota units per month and does not require a card. A no-account citation check is also available at vantagemcp.dev/check, but that page is separate from connecting an MCP client.

For a local installation, install the repository in editable mode with uv pip install -e ., set DATAFORSEO_USERNAME and DATAFORSEO_PASSWORD, and run python -m vantage_mcp.server. Add --http to use HTTP mode and set VANTAGE_PORT if a specific port is needed; the README gives 8420 as an example.

Tools and capabilities

The Vantage MCP server provides these operations:

• get_usage checks remaining quota without spending units.
• check_prompt_coverage tests up to 10 keywords for citations to a domain and can separately inspect brand mentions.
• find_citation_leaders ranks domains cited for a topic and checks whether a chosen domain appears.
• analyze_citation_trend returns month-by-month mention counts for a domain.
• analyze_citation_structure describes the format of a winning cited answer.
• analyze_citation_structure_batch compares that format across up to 10 related topics.
• analyze_citation_gap compares a page with the winning AI-cited answer for the same keyword.

Limitations and notes

The server measures sampled AI-answer citation behavior; it does not guarantee that changing a page will result in future citations. Quota applies to metered operations, and the old check_ai_visibility operation is retained but documented as deprecated. Self-hosting is not a complete billing or account-management system, and requires external DataForSEO credentials. The project is licensed under MIT.