What CorpusIQ MCP server does
CorpusIQ MCP server gives AI assistants access to live data from business systems through the Model Context Protocol. The service is aimed at operators, analysts, accountants, agencies, founders, and autonomous agents that need answers from operational data rather than static reports.
The repository describes more than 40 connectors and a listing of 65+ MCP tools. Available data sources include Shopify, Stripe, Amazon Seller, eBay, QuickBooks, HubSpot, Google Analytics 4, Google Ads, Meta Ads, Klaviyo, Slack, Gmail, PostgreSQL, MongoDB, Google Drive, Notion, and other supported services. Connector coverage spans commerce, marketing, CRM, finance, email and SMS, communication, SEO, databases, and files.
A typical request could compare Shopify orders with Stripe charges, inspect QuickBooks financial data, or review marketing results across several advertising platforms. The returned source data includes citations, while the requesting AI client's conversation policy applies after the data is received.
How it works
CorpusIQ MCP server uses the Model Context Protocol and a streamable HTTP transport. The documented endpoint is:
https://mcp2.corpusiq.io/mcp
The client configuration identifies the server URL and sets the transport to streamable-http. After the connection is configured, an MCP-compatible assistant can invoke the available data tools. CorpusIQ retrieves information from the connected providers and passes it to the AI client.
The documented workflow has three stages: authorize retrieval from each external provider, ask a question in an AI assistant, and receive an answer based on the retrieved data. This makes the service suitable for cross-source questions that would otherwise require several exports or separate reports.
Setup and configuration
The documentation directs users to create an account, connect an AI client, and run a first query. Each external source is authorized through that provider's documented OAuth flow. Connector-specific setup guides are available for the supported integrations.
A compatible client needs an MCP server entry similar to this:
{
"mcpServers": {
"corpusiq": {
"url": "https://mcp2.corpusiq.io/mcp",
"transport": "streamable-http"
}
}
}
The provided material does not specify client-side environment variables, local package installation, or a self-hosted deployment method. The repository is primarily documentation for the hosted service and its connectors.
Tools and capabilities
CorpusIQ MCP server supports capabilities including:
- Querying live commerce data from platforms such as Shopify and Stripe
- Retrieving finance information including QuickBooks profit and loss, invoices, and balance sheets
- Examining marketing, advertising, analytics, and SEO data
- Accessing CRM, email, SMS, communication, database, and file sources
- Combining information from multiple connected sources in one question
- Returning source citations with retrieved data
- Supporting AI assistants and autonomous agents through MCP
The repository also includes connector guides, API documentation, troubleshooting material, recipes, and prompt examples. These resources explain setup and usage but do not establish that every connector supports write operations.
Limitations and notes
Access depends on connecting and authorizing the relevant external providers. Connector availability and supported query coverage may differ by source; the provided material does not enumerate the exact operations for every connector.
CorpusIQ is presented as a hosted HTTP service rather than a package to install from this repository. The README does not state pricing, service limits in detail, data-retention terms, or whether all listed connectors are available on the same account plan. Review the service documentation for current connector, rate-limit, OAuth, and account requirements before deployment.
The repository is licensed under MIT and is open for documentation contributions. Its contents include instructions and reference material for the CorpusIQ service, not an implementation of a local MCP server.