Hosted vs local MCP
Compare Pear's hosted MCP endpoint with a separate third-party or custom local MCP server across setup, data paths, credentials, maintenance, client fit, and cost.
A local MCP client is not the same thing as a local MCP server. The client is the AI application that connects to servers; each server can run on your machine or at a remote endpoint. A local client can therefore call hosted Pear when it supports remote MCP and a documented Pear authentication path.
The AI application runs on your machine. It can still connect to remote MCP servers, cloud models, and external providers.
A separate third-party or custom MCP server runs in infrastructure you control. You own its credentials, updates, availability, and provider integration.
The MCP server runs as a remote service. Pear is in this category and exposes one Streamable HTTP endpoint.
Neither model is automatically private, secure, available, or inexpensive. The practical difference is who operates each part and which systems receive the request.
Hosted Pear
Pear operates the remote MCP service at https://pearmcp.com/api/mcp. You configure your Pear account, provider connections, and AI client.
Separate local MCP server
You independently select or build an MCP server, run it on infrastructure you control, and choose how the AI client reaches it.
Hosted Pear
Create a Pear account, connect an available provider, then follow the maintained setup guide for your AI client.
Separate local MCP server
Install and configure the server, its runtime, provider integration, credential storage, transport, and client entry yourself.
Hosted Pear
The endpoint is independent of your personal computer, but requests still depend on Pear, your connected provider, the network, and the AI client. Pear does not promise uninterrupted availability.
Separate local MCP server
Availability depends on the machine, process supervision, power, networking, and provider integration you operate.
Hosted Pear
The AI client sends Pear a structured tool call and its arguments, not the full chat transcript. Pear fetches provider data live and returns the result; approved-write records and operational metadata can be retained as described on the data-handling page.
Separate local MCP server
The tool call reaches the local server first, but the server still contacts the provider. Your AI client, model service, provider, logs, and any other configured systems keep their own data paths, so local does not mean all data stays on the device.
Hosted Pear
Pear stores encrypted provider credentials or OAuth tokens and decrypts them in memory when a provider request needs them. AI-client access uses the authentication path documented for that client.
Separate local MCP server
You choose and maintain the local credential store, file permissions, authentication, secret injection, and revocation process. The exact boundary depends on the server you select.
Hosted Pear
Pear maintains the deployed service and its provider integrations. You still manage provider permissions, client configuration, account access, and any approval choices.
Separate local MCP server
You own installation, updates, dependency and operating-system patches, logs, backups, process restarts, network exposure, and provider breakages.
Hosted Pear
Pear publishes setup paths for named AI clients and currently provides Apple iCloud and Microsoft 365 tools. Client plans, roles, permissions, and current host behavior can still limit what is available.
Separate local MCP server
Compatibility and provider coverage depend entirely on the local server and client you choose. A project listing a provider or client is not proof that its current auth and tool paths work for your setup.
Hosted Pear
Pear has a Free plan and optional paid plans. The pricing page and Stripe Checkout are the current sources for limits, eligibility, price, and currency.
Separate local MCP server
A local server may be free or paid, while compute, hosting, maintenance time, monitoring, and support remain your responsibility. Check the selected project and infrastructure directly.
Pear does not recommend or certify a third-party local server on this page.
An operator can configure Pear and a separate third-party or custom local MCP server in the same AI host when that host supports multiple servers. That is an operator-chosen stack across separate products and trust boundaries, not a hybrid Pear mode.
Pear is a closed-source hosted service available through https://pearmcp.com/api/mcp. Pear does not provide source code, a local server download, or a self-hosted deployment. Local AI clients can call the hosted endpoint when their remote transport and authentication path are supported.
Product boundary reviewed
Current provider boundary
Pear supports Apple iCloud and Microsoft 365 today. Compatible IMAP/SMTP mailbox support is a beta when enabled. Google Calendar, Gmail, Google Tasks, and Google Contacts are coming soon.
https://pearmcp.com/api/mcpApple Reminders boundary
Pear's hosted CalDAV connection cannot access Apple's upgraded Reminders store. Microsoft To Do uses Microsoft Graph and is unaffected.
Apple Reminders availability depends on the Apple account. For Apple, Pear can read and write only the legacy reminder lists Apple still exposes through CalDAV. If the account has been upgraded to Apple's newer Reminders store, Pear cannot access those native lists and does not copy them into a separate reminder store. Apple may instead return compatibility placeholders, such as repeated "Where are my reminders?" items; Pear did not create them. Pear's safe fallback hides known placeholders and blocks writes to affected lists. Reconnecting iCloud or replacing the App-Specific Password will not change this Apple limitation. Calendar, Contacts, and Mail are unaffected, and Microsoft To Do uses Microsoft Graph and is unaffected. Full modern Apple Reminders access would require an Apple-platform companion using EventKit with the user's permission; Pear does not currently provide one.
Read the Apple Reminders compatibility noteThese answers describe Pear's current product boundary, not every MCP server or AI client.
Pear is a closed-source hosted service available through https://pearmcp.com/api/mcp. Pear does not provide source code, a local server download, or a self-hosted deployment. Local AI clients can call the hosted endpoint when their remote transport and authentication path are supported.
Yes, when the client can call a remote Streamable HTTP endpoint and use a Pear authentication path documented for that client. A setup guide is a documented path, not proof that every client version, plan, workspace policy, role, permission, or write flow is available.
No. A local server changes where that server process runs. The AI client or model service, connected provider, update services, logs, and any other systems you configure can still receive data. Review the complete path for the exact server and client you choose.
Pear receives the structured MCP tool call and arguments the AI client chooses to send, then calls the connected provider. Pear does not receive the full chat transcript. Ordinary reads are fetched live rather than copied into a synced provider database; approved-write records and bounded operational metadata have separate retention described on the data-handling page.
Start Free, connect an available provider, and follow your AI client’s setup guide.