Frequently Asked Questions

Q: What applications and platforms support MCP integrations?

A: The Model Context Protocol is rapidly gaining adoption where each client supports different MCP features (tools, prompts, resources) depending on their use cases. For the semi-complete list of Clients with feature compatibility details, visit the official MCP clients directory.

Q: What Claude plan do I need to use MCP integrations?

A: MCP integrations, including the Grain MCP server, are only available on paid Claude plans (Pro, Team, and Enterprise). Free Claude accounts do not support MCP integrations or the ability to connect external tools and data sources. You'll need to upgrade to a paid plan to access the Grain integration and other MCP servers.

Q: Are all Grain MCP tools available on every Grain plan?

A: Most Grain MCP tools work on all Grain plans (Free, Starter, Business, Enterprise), including meeting search, transcripts, notes, and contact management. However, tools for Deals (list_open_deals, list_all_deals, fetch_deal) and Sales Coaching/Scorecards (list_coaching_feedback, fetch_meeting_coaching_feedback) are restricted to Business and Enterprise plan users only. Free and Starter plan users will receive an error if they attempt to use these premium tools.

Q: Is it safe to use MCP servers like Grain's?

A: MCP security relies on a trust model, and Grain's MCP server is officially supported by us and aligns with our existing security practices, follows the MCP specification exactly, and uses industry-standard OAuth authentication and HTTPS. All tool usage requires explicit user approval in Claude, and each tool request is scoped to specific data needed for that operation. While no system is 100% secure, Grain's established security practices and transparent approach to data handling make it a safer MCP implementation.

Q: How does the Model Context Protocol (MCP) work and what are the privacy implications?

A: The Model Context Protocol (MCP) is an open standard that creates a secure bridge between Claude and external services like Grain, where your data stays in your Grain account and is only processed temporarily by Claude to answer questions. As stated by Anthropic, data accessed through MCP integrations is not used to train Claude's models.

Q: What data does Grain's MCP server have access to?

A: Grain's MCP server only accesses data you already have permission to view in your Grain workspace, including meetings you attended, companies and contacts from those meetings, deal information (with HubSpot integration), and AI-generated notes and transcripts. It has no access to private meetings you weren't invited to or other workspaces you don't belong to. The integration uses OAuth authentication and maintains the same security boundaries as your normal Grain usage.

Q: How does Claude verify and approve MCP tool usage?

A: Claude requires explicit user approval before using any Grain tool, showing you a dialog with clear descriptions of what each tool will do and offering "use once" or "use for entire session" options. Each tool requires separate approval when first used, and you maintain control over which tools Claude can access in your current session. You can revoke access by starting a new chat session, ensuring that no tools run without your explicit permission.

Q: Can I control what data the Grain MCP server shares with Claude?

A: Data access is controlled through your existing Grain account permissions, and you can enable/disable specific tools in Claude's integration settings, but you cannot currently exclude specific meetings or companies from MCP access. The integration accesses all data you have permission to see in your Grain workspace and follows your existing permissions rather than providing additional filtering options. Review your Grain workspace permissions before connecting and use targeted prompts to limit the scope of data Claude requests.

Q: What is the helpful_usage_information parameter I see in the client's tool responses?

A: You may notice the helpful_usage_information parameter appearing in chat interface when the Grain MCP server responds to tool requests. This optional parameter helps Grain's team understand tool usage patterns to improve functionality, and this data is never used as AI training data. Currently there's no opt-out option, but one will be available in the coming months - if you're uncomfortable with this data collection, consider waiting until then.