NexusFeed MCP is a data access service for AI agents that provides structured, verifiable JSON data through the Model Context Protocol. Its core purpose is to provide AI agents with two types of real-time compliance and business data that cannot be obtained directly from raw web pages, enabling agents to assess data quality before compliance-critical decisions.
Main capabilities include two types of datasets: first, real-time LTL (less-than-truckload) freight fuel surcharge rates, with data extracted from carrier rate pages; second, ABC liquor license compliance records for U.S. states, covering California (CA), Texas (TX), New York (NY), and Florida (FL), supporting searches by trade name, owner, or license number, and verifying license status, expiration date, and address. These data sources typically involve JavaScript rendering, CAPTCHAs, and session state, which raw LLM browsing methods are structurally unable to access. Each tool response includes a _verifiability block, which includes an extraction timestamp, confidence score, and source URL, enabling agents to assess data quality before acting.
Regarding usage, users need to subscribe through RapidAPI to obtain an X-API-Key. A Freemium plan is offered, with 10 free requests per day. Configuration supports multiple MCP clients: Claude Desktop can be configured to use only LTL fuel surcharge rates or only ABC license compliance data; Cline (VS Code) allows manually adding the server under MCP Servers in settings; it can also be used via Streamable HTTP (Smithery or remote clients), requiring the X-API-Key request header to be passed with each request. Server metadata (no authentication required) is also accessible.
In actual use, tool responses provide three data quality judgment fields: extraction_confidence must be greater than or equal to 0.90 before it can be used for compliance-critical decisions; if source_timestamp is within the data_freshness_ttl_seconds range, it indicates the data comes from cache and is fresh; raw_data_evidence_url is the canonical source, which agents can independently verify. Typical scenarios include auditing LTL freight invoices and verifying liquor licenses before distributor transactions.
Regarding code and licensing: this client code (this repository) is licensed under the MIT License; the NexusFeed backend service (https://api.nexusfeed.dev）为商业服务，上述 The MIT License covers only the Python client wrapper in this repository and does not grant rights to the data service, which requires a paid API key and is subject to separate terms of service. Enterprise-grade SLAs and licensing can be obtained by contacting ops@nexusfeed.dev.
Limitations: Detailed data field specifications, support for states beyond the covered state scope, response times, rates, and other specific parameters are not currently provided.
