Fullmention - Track how AI recommends your brand

What is Fullmention - Track how AI recommends your brand?

Fullmention is a brand visibility monitoring service that tracks how AI models (like ChatGPT and Gemini) recommend your brand. The Fullmention MCP (Model Context Protocol) Server acts as a secure gateway, allowing AI agents such as Claude Desktop, Cursor, or custom agent networks to interact directly with the Fullmention Public API.

How to use Fullmention - Track how AI recommends your brand?

Install the MCP server via npx with zero installation. For Claude Desktop, add a configuration entry to claude_desktop_config.json specifying the npx command and your FULLMENTION_API_KEY. For Cursor, add a new MCP server in Settings → Features → MCP with the same command and environment variable. The server exposes 17 dedicated MCP tools and structured resources for keyword management, batch analysis, and analytics.

Key features of Fullmention - Track how AI recommends your brand

• In-memory query shielding with 60-second TTL to protect API rate limits
• Status polling guardrails to prevent rapid-succession polling of active runs
• Conflict safety for batch runs, extracting active runId without crashing
• Server-side Share of Voice calculation to eliminate LLM hallucinations
• Token-optimized Markdown format to reduce LLM input costs
• SSE (Server-Sent Events) mode for cloud deployment and diagnostics endpoint
• Read-only security mode for public or shared team environments

Use cases of Fullmention - Track how AI recommends your brand

• Monitor how AI search engines recommend your brand across keywords and countries
• Manage keyword portfolios – create, update, delete, and list keywords with tags
• Estimate credit consumption before triggering batch analysis runs
• Trigger and track batch analysis runs (including web search fanout)
• Calculate brand Share of Voice percentage and average placement rankings

FAQ from Fullmention - Track how AI recommends your brand

What does the Fullmention MCP Server do?

It exposes Fullmention’s recommendation snapshots as native MCP tools and resources, enabling AI models to monitor AI visibility, manage keywords, estimate costs, and trigger batch analysis runs without writing custom API code.

Does the server handle API rate limits and quotas automatically?

Yes. It caches query responses (60-second TTL) to protect external API rate limits from redundant agent loops and includes status polling guardrails that intercept rapid polling within 8 seconds, returning cached state with a warning note.

Can I deploy the server in a read-only mode?

Yes. Set FULLMENTION_MCP_READONLY=true when launching in SSE mode to block all mutating actions (e.g., triggering runs, creating/deleting keywords), suitable for public gateways or shared team environments.

Are there any known issues or limitations?

Filtering results or list endpoints strictly by engine (especially gemini) can occasionally return a 500 Internal Server Error from the backend API because Gemini does not support web search fanout.

How does tagging work with Fullmention?

Tags use a prefix-based hierarchy (e.g., client:acme, market:dk). Queries require the fully qualified, prefixed tag name (e.g., client:acme not just acme) to match correctly.