Review your training
Get a summary of the last 1–12 weeks. Look up individual workouts, past sets for an exercise, or changes in body measurements.
“How consistent have I been over the last four weeks?”
Ask about
your Hevy
workouts.
Hevy MCP connects Claude, Codex, and other AI assistants to your Hevy account. Ask about past workouts, create routines, or update your log.
Open source · Hevy PRO API key required
TRY ASKINGEXAMPLE
hevy / get-training-summary 6 weeks
Example request. Connect your account to get your own results.
WORKS WITH
Claude / Codex / Cursor / Other MCP clients
What you can do
The server has 22 tools for reading and updating your Hevy data.
Create and edit routines
Find a saved routine or create a new one. Choose exercises, sets, rep ranges, and rest times through your assistant.
“Find my push-day routine and show me its exercises and sets.”
Log completed workouts
Save a workout after training or correct an existing entry. You can also add custom exercises and record body measurements.
“Log a workout from my routine. Ask me for the actual set results first.”
How it works
MCP stands for Model Context Protocol. It lets an AI assistant connect to services such as Hevy. This server uses your Hevy API key to read or update your account when the assistant requests it.
You ask your assistant
Ask a question about your workouts, or describe what you want to create or change.
The assistant calls Hevy MCP
The server sends the request to Hevy. You can use the hosted server or run it on your computer.
Your assistant gets the result
It uses the returned data to answer your question or tell you whether the requested update succeeded.
Some tools change your Hevy data. If your assistant supports tool approvals, use them to check changes before they are saved.
From the community
Comments from Reddit. Follow the links to read each discussion in full.
“I added it to my hermes setup and it works great to get an overview of my workouts. No issues so far”
“Have been using this one for weeks, it's pretty decent”
Replying to a link to chrisdoc/hevy-mcp.
“I use this multiple times a week and it's been super helpful. I use it to analyze my training on a weekly basis, create new training blocks and routines.”
The full comment also raises concerns about telemetry, self-hosting, tool size, and updates.
“yes I start using now this https://github.com/chrisdoc/hevy-mcp with Cursor and is much better.”
Posted in a discussion about problems with an earlier ChatGPT integration.
“I asked Claude code to find one 2 days ago, it reported it found 8 existing ones already, it did the analysis and selected the best and most mature one on github and installed it (chrisdoc/hevy-mcp on github).”
Describing Claude Code’s selection. The commenter went on to use it to modify their training programme.
“Nice work dude, this is awesome!”
A reaction to the project announcement.
Set up Hevy MCP
You’ll need a Hevy API key and an assistant that supports MCP. Use the hosted server to connect without installing anything, or run it on your own computer.
1
Get your Hevy API key
API access requires a Hevy PRO subscription. Create your key in Hevy’s API settings ↗ and keep it secure. You’ll enter it during setup, not on this website.
2
Connect your assistant
Set up Claude
NO INSTALL REQUIREDIn Claude, open Settings → Connectors → Add custom connector. Paste this URL:
https://mcp.hevy-mcp.dev/mcpFollow the authorization steps and enter your Hevy API key when asked. Once connected, try the question below.
Your Claude account needs access to custom connectors. The same URL works in Claude Desktop.
Set up Codex
In the terminal you use to start Codex, set your API key and add the server:
export HEVY_API_KEY=your-hevy-api-key
codex mcp add hevy \
--url https://mcp.hevy-mcp.dev/mcp \
--bearer-token-env-var HEVY_API_KEYReplace your-hevy-api-key with your key. Restart Codex or start a new session, then run codex mcp list to check the configuration.
Codex CLI, the desktop app, and the IDE extension share this configuration. The app you launch must have access to the HEVY_API_KEY environment variable.
Set up another HTTP client
Your client needs to support Streamable HTTP and an Authorization header. Use this as a starting point and check your client’s documentation for its configuration format.
{
"mcpServers": {
"hevy": {
"url": "https://mcp.hevy-mcp.dev/mcp",
"headers": {
"Authorization": "Bearer your-hevy-api-key"
}
}
}
}Replace your-hevy-api-key with your key and reconnect your client. Keep the configuration file private because it contains your key.
Run on your own machine
Install Node.js 20 or newer, then add the entry below to your client’s MCP configuration. This works with Claude Desktop, Cursor, and other clients that can start a local MCP server. Keep any existing server entries.
{
"mcpServers": {
"hevy": {
"command": "npx",
"args": ["-y", "hevy-mcp"],
"env": {
"HEVY_API_KEY": "your-hevy-api-key"
}
}
}
}Replace your-hevy-api-key with your key, keep the file private, and restart your client. Cursor uses ~/.cursor/mcp.json.
3
Ask your first question
Give me a training summary for the last four weeks.
If the connection fails, check that your API key is correct and that your client supports the setup method you chose. Try restarting the client after saving its settings. Read the setup docs ↗
What happens to your data
The hosted server processes the Hevy data needed to answer each request. It doesn’t keep a shared database of your workouts. Keys sent directly in an Authorization header aren’t saved. For OAuth connections, authorization grants are stored encrypted so your assistant can stay connected.
The results are sent to your AI assistant, whose own privacy policy applies. The hosted server also collects operational data, described in the privacy policy below. If you run Hevy MCP locally, requests go from your computer to Hevy.
Read the privacy policy ↗