hevy-mcp
Exposes the Hevy workout API to Claude, enabling users to manage workouts, routines, exercise templates, body measurements, and user info via natural language.
README
Hevy MCP Server
A single-user remote MCP server that exposes the
Hevy workout API to Claude (and any OAuth-capable MCP client).
It speaks the full MCP protocol over Streamable HTTP at /mcp and is protected by a minimal,
stateless OAuth 2.1 flow, so it can be added as a custom connector in the Claude app —
including on mobile, on the go.
- Runtime: Bun or Cloudflare Workers — the same code runs on both, via Hono +
@hono/mcp. - Auth: single-user OAuth 2.1 (PKCE S256), all state held in signed JWTs — no database.
- Hevy: a single
api-keyfrom an environment variable / secret.
Why OAuth (and not just a token)?
As of 2026 the Claude app's custom-connector UI only accepts OAuth — it does not let
you paste a static bearer token or custom header. So this server implements a tiny
single-user OAuth provider: when you connect, Claude sends you to a login page where you
enter MCP_PASSWORD; on success it receives an access token (a signed JWT) that it then
presents to /mcp. Nothing is persisted — the password gates the flow and PKCE + JWT
signatures secure it. Because the server supports Dynamic Client Registration, you do not
need to fill in any Client ID / Client Secret in Claude — just the server URL.
Prerequisites
- A Hevy account with API access — the developer API key lives in the Hevy app under Settings → Developer and requires a Hevy Pro subscription.
- For local development / Bun deploys: Bun ≥ 1.1 (
curl -fsSL https://bun.sh/install | bash). - For Cloudflare deploys: a Cloudflare account and Node.js ≥ 22 (Wrangler requires it). Bun is not needed on Workers.
- A public HTTPS URL to deploy to (Coolify/Docker host or Cloudflare). The Claude app
connector requires HTTPS —
localhostworks only for local testing with the MCP Inspector.
Tools
All ~22 Hevy endpoints, read + write:
- Workouts:
list_workouts,get_workout,get_workout_count,get_workout_events,create_workout,update_workout - Routines:
list_routines,get_routine,create_routine,update_routine - Routine folders:
list_routine_folders,get_routine_folder,create_routine_folder - Exercise templates:
list_exercise_templates,get_exercise_template,create_exercise_template - Exercise history:
get_exercise_history - Body measurements:
list_body_measurements,get_body_measurement,create_body_measurement,update_body_measurement - User:
get_user_info
Configuration
| Variable | Description |
|---|---|
HEVY_API_KEY |
Your Hevy API key (Hevy app → Settings → Developer; requires Hevy Pro). |
MCP_PASSWORD |
The password you type on the OAuth login page when connecting Claude. |
JWT_SECRET |
Random ≥32-char secret for signing JWTs. Generate with openssl rand -hex 32. |
PUBLIC_URL |
Public HTTPS base URL (no trailing slash). Must match the deployed domain. |
PORT |
Port to listen on (default 3000). Ignored on Cloudflare Workers. |
Run locally (Bun)
bun install
cp .env.example .env # then edit values
bun run dev # or: bun run start
Then point an MCP client at http://localhost:3000/mcp, e.g. the inspector:
npx @modelcontextprotocol/inspector
It discovers the OAuth metadata, opens the login page (enter MCP_PASSWORD), and then lets
you list and call tools.
Deploy: Docker (Coolify or any host)
The included Dockerfile runs on oven/bun and has a built-in /health healthcheck.
Generic Docker:
docker build -t hevy-mcp .
docker run -p 3000:3000 \
-e HEVY_API_KEY=... \
-e MCP_PASSWORD=... \
-e JWT_SECRET="$(openssl rand -hex 32)" \
-e PUBLIC_URL=https://hevy-mcp.example.com \
hevy-mcp
Put it behind a reverse proxy (Caddy/nginx/Traefik) that terminates TLS, and set
PUBLIC_URL to the public HTTPS address.
Coolify:
- Create a new resource from this repo (Dockerfile build). Exposed port:
3000. - Set the environment variables above. Set
PUBLIC_URLto the domain Coolify assigns after it's assigned (Traefik provisions TLS automatically) — it must match exactly. - Health check path:
/health. - Deploy.
Deploy: Cloudflare Workers
The same entry point runs on Workers (env comes from the request binding, not process.env;
nodejs_compat is enabled in wrangler.jsonc).
-
Install deps and log in (needs Node.js ≥ 22):
bun install # or: npm install npx wrangler login -
Set
PUBLIC_URLinwrangler.jsoncto your Worker's URL. Deploy once to discover it (e.g.https://hevy-mcp.<your-subdomain>.workers.dev), or use a custom domain / route, then updatePUBLIC_URLto match and redeploy. -
Add the three secrets:
npx wrangler secret put HEVY_API_KEY npx wrangler secret put MCP_PASSWORD npx wrangler secret put JWT_SECRET # e.g. paste `openssl rand -hex 32` -
Deploy:
bun run cf:deploy # = wrangler deployLocal dev against the real Workers runtime: copy
.dev.vars.exampleto.dev.vars, setPUBLIC_URL=http://localhost:8787there, thenbun run cf:dev.
Connect from Claude
In the Claude app, add a custom connector pointing at https://<your-domain>/mcp. Leave
Client ID / Client Secret blank. Complete the OAuth prompt by entering MCP_PASSWORD, and
the Hevy tools become available.
Security notes
This repo is safe to keep public: no secrets live in it. Access is guarded entirely by two
values you set as env vars / secrets — use a strong MCP_PASSWORD and keep JWT_SECRET
private (rotating JWT_SECRET invalidates existing tokens and forces re-authentication).
Implementation notes
- The server runs statelessly — a fresh MCP server + transport per request — which is
simplest and robust for a single user. It does not advertise a server→client SSE stream
(GET
/mcp); tool calls are request/response. - Authorization codes are short-lived (5 min) signed JWTs and are not single-use within that window — an acceptable trade-off for a private, single-user server with no datastore.
License
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.