Connect Hevy to an AI assistant
Hevy MCP lets an AI assistant access workout data from Hevy. It uses the Model Context Protocol (MCP), a standard that lets AI applications use external tools and data.
What you can do
- Ask about workout history and routines without copying them into a chat.
- Find exercise templates and inspect a training plan.
- Create or update supported Hevy routines.
Hevy remains the source of workout records; the assistant explains the data it retrieves. The example above uses synthetic data. It shows a read request, not a live connection to a personal account.
The same operations run locally and in the cloud
The repository separates the Hevy client, reusable domain operations, and MCP tools from the runtime adapters. A Node.js adapter serves local clients over standard input and output. A Cloudflare Workers adapter serves hosted connections over HTTP.
That separation lets the adapters reuse the same operations while handling their own connection and authentication requirements. It also means the shared code needs to respect the capabilities of both runtimes.
What it cannot guarantee
MCP connects the assistant to Hevy; it does not guarantee that the assistant interprets a workout correctly. Its answers depend on which records are available and how it uses them.
The connection requires Hevy API access. Reading and changing data are different actions. Review a proposed routine or update before applying it. Supported tools and setup requirements can change, so check the product documentation for current instructions.
Explore the implementation
The source repository contains the current implementation and tool reference. For connection options and setup, head to hevy-mcp.dev.