Whalesync

What Whalesync MCP server does

The Whalesync MCP server gives AI agents access to Whalesync’s sync management API. Whalesync keeps data synchronized in both directions between supported tools, including Airtable, Webflow, HubSpot, Salesforce, Notion, and Postgres. The agent can create a sync between two apps, inspect available data structures, define how fields correspond, enable or pause the sync, and review synchronization problems.

The tool set contains 24 operations grouped around building, running, monitoring, fixing, and navigating syncs. Build operations cover sync creation and updates, connector discovery, base/table/field listing, mapping retrieval and changes, and mapping validation. Runtime operations activate, pause, or delete a sync. Monitoring operations expose status, issues, operations, record searches, record status, and pending deletes.

How it works

The service is hosted remotely at https://api.whalesync.com/mcp and communicates through Streamable HTTP. An MCP-compatible client connects to that URL and invokes tools with the sync_ prefix. The prefix identifies tools related to Whalesync’s sync functionality.

Authentication uses OAuth 2.1 with PKCE. On first use, the client opens a browser where the user signs in to Whalesync and grants access. Consent can be limited to read-only access. A Whalesync account is required.

Actions that can have significant consequences remain outside autonomous tool execution. Connecting apps through OAuth, starting a sync, and approving deletes take place in the Whalesync application. The agent provides a link for the user to complete those steps.

Setup and configuration

The Whalesync MCP server requires no local installation from this repository. Add the remote endpoint to a supported client using its HTTP MCP configuration, or use the client’s installation flow where available. The repository is a public reference point for the server; the running service is hosted by Whalesync.

The README provides setup paths for Cursor, Claude Code, VS Code, Codex, Claude, ChatGPT, and Grok. For clients that accept a direct URL, use https://api.whalesync.com/mcp. Cursor, Claude Code, VS Code, and Codex each have documented commands or configuration examples in the project README.

After connection, complete the browser-based Whalesync authorization. No API key needs to be copied into client configuration.

Tools and capabilities

The Whalesync MCP server supports these workflows:

• Discover connectors and inspect available bases, tables, and fields.
• Create or update sync definitions between apps.
• Read, modify, and validate table and field mappings.
• List and inspect existing syncs.
• Activate, pause, or delete syncs, subject to Whalesync’s human approval flow.
• Check sync status and enumerate operations and unresolved issues.
• Inspect individual issues, retry failed issues, and search records.
• Review a record’s sync status and identify pending deletes.
• Use a “what’s next” navigation operation to continue working with syncs.

These capabilities are suited to assistants that need to translate a natural-language synchronization request into configuration steps, then help diagnose why a record has not synchronized.

Limitations and notes

The Whalesync MCP server is remote-only. Nothing in this repository installs a local server, and the server itself is developed elsewhere. Availability and behavior therefore depend on the hosted Whalesync endpoint and the user’s account permissions.

The available tools focus on Whalesync syncs rather than direct arbitrary data editing in connected applications. Human approval is still required for app connections, sync activation, and delete approval. The README identifies Airtable, Webflow, HubSpot, Salesforce, Notion, and Postgres as examples of connected tools, but does not provide a complete connector list here.

Documentation is available at https://docs.whalesync.com/api/mcp; support is provided at support@whalesync.com.