Visibility Index

What visibility-index MCP server does

The visibility-index MCP server makes the AI Visibility Index queryable through the Model Context Protocol. It connects assistants to two published data desks: dabyte.ai, which covers SaaS and AI tools, and dablock.ai, which covers crypto and Web3. The available datasets contain current measurements, historical time series, and derived rankings for the brands tracked in each niche.

The measurements describe share of answer: the percentage of a fixed set of category-buyer prompts in which an answer engine names a brand. The index measures ChatGPT, Perplexity, and Google Gemini. Each prompt is run against each engine weekly, and the published records include overall and per-engine shares, ranks, commercial-intent scores, quadrants, and panel versions.

How it works

Hosted MCP endpoints are available at https://dabyte.ai/mcp for SaaS and AI tools and https://dablock.ai/mcp for crypto and Web3. The server reads the published JSON over HTTPS, so a self-hosted instance does not need a local copy of the dataset. The repository mirror is intended for discovery and reproducibility; the two domains remain the canonical sources for current data.

The underlying panel contains 16 prompts per niche and is frozen between releases. Changes to the prompt panel create a new version, and the methodology does not compare deltas across panel versions. Past measurements are archived at permanent URLs, allowing published changes to be recomputed.

Setup and configuration

For hosted use, configure an MCP client with the endpoint for the niche you need. No API key or account is required according to the published material.

To run the server locally, build and start the included Docker image:

docker build -t aiv-mcp .
docker run -p 8090:8090 aiv-mcp

The container reads the published JSON over HTTPS and does not require the repository’s mirrored dataset. The repository also includes scripts/fetch_latest.py for refreshing the local data mirror, but the README does not state that this script is required by the MCP server.

Tools and capabilities

The visibility-index MCP server supports assistant queries against the published index data, including:

• Current per-brand visibility from aiv.json.
• Historical brand measurements from history.json.
• Derived rankings from rankings.json.
• Overall and engine-specific results for ChatGPT, Perplexity, and Gemini.
• Brand rank, commercial-intent score, quadrant, and panel-version fields.
• Separate SaaS and AI-tool or crypto and Web3 datasets.

The repository does not provide individual MCP tool names in the supplied material, so client-side tool names should be confirmed in mcp-server/README.md.

Limitations and notes

The index measures whether a brand is named in a defined prompt panel; it is not a general measure of brand awareness or all assistant responses. Results are constrained by the published panel, release schedule, engines measured, and measurement resolution. Small movements within one scale step are not reported as changes.

The repository contains two license regimes. Dataset files under data/ use CC BY 4.0 and require attribution, while the code in scripts/ and mcp-server/ uses MIT. The README does not identify specific MCP client applications or provide client-specific setup instructions. Placement is not sold, and records include an is_client field to indicate whether a brand is a client.