MCP Connector

Connect LLMs to Strava Data

Access Strava activities, athlete stats, segments, and routes - view heart rate/power data and export routes as GPX/TCX.

Works with strava

90
Spark score
out of 100
Updated last month
Version 1.0.0
Models

Add to Favorites

Why it matters

Enable Large Language Models to access and analyze your Strava activity data, athlete profiles, segments, and routes. This MCP server facilitates fitness tracking and data-driven insights.

Outcomes

What it gets done

01

Retrieve recent activities, athlete profiles, and statistics.

02

Access detailed activity data streams (power, heart rate, cadence).

03

Manage and explore Strava segments and routes.

04

Export route data in GPX or TCX formats.

Install

Add it to your toolbox

Run in your project directory:

curl -fsSL https://spark.entire.vc/get/vb-strava | bash

Capabilities

Tools your agent gets

get-recent-activities

Retrieves recent activities for the authenticated user

get-athlete-profile

Retrieves profile information for the authenticated athlete

get-athlete-stats

Retrieves activity statistics (recent, year-to-date, all-time) for the authenticated athlete

get-activity-details

Retrieves detailed information about a specific activity by its ID

list-athlete-clubs

Lists the clubs the authenticated athlete is a member of

list-starred-segments

Lists segments starred by the authenticated athlete

get-segment

Retrieves detailed information about a specific segment by its ID

explore-segments

Searches for popular segments in a given geographic area (bounding box)

+4 tools

Overview

Strava MCP Server

Strava MCP Server gives Claude access to Strava activities, athlete stats, segments, and routes via Strava API V3, with automatic token refresh and route export to GPX/TCX. Use it for conversational access to Strava fitness data. Requires completing the Strava OAuth setup with the callback domain set to localhost before use.

What it does

Strava MCP Server integrates the Strava API with an LLM, giving it access to recent activities, athlete profile and stats, segments, segment efforts, clubs, and saved routes for fitness data tracking and analysis, using Strava API V3 with automatic token refresh.

When to use - and when NOT to

Use it when you want to ask things like "Show me my recent Strava activities," "What were my last 3 rides?," "Get the heart rate data from my morning run yesterday," or "Show me the power data from my last ride." It also supports exploring and starring segments, listing your segment efforts and clubs, and exporting saved routes to GPX or TCX files on your local filesystem. Do not skip the exact setup sequence - the project specifically requires: install the server, configure it in Claude, complete Strava's OAuth authorization, then restart Claude - and the Strava app's authorization callback domain must be set to localhost.

Capabilities

  • get-recent-activities, get-activity-details: recent activities and detail for a specific activity (including power/heart rate/cadence streams).
  • get-athlete-profile, get-athlete-stats: profile info and recent/YTD/all-time statistics.
  • list-athlete-clubs: clubs the athlete belongs to.
  • list-starred-segments, get-segment, explore-segments (search by geographic bounding box), star-segment (star/unstar).
  • get-segment-effort, list-segment-efforts (with optional date filtering).
  • list-athlete-routes: list saved routes, with export to GPX/TCX format.

How to install

git clone https://github.com/r-huijts/strava-mcp.git
cd strava-mcp
npm install
npm run build

Configure Claude Desktop:

{
  "mcpServers": {
    "strava-mcp-local": {
      "command": "node",
      "args": ["/absolute/path/to/your/strava-mcp/dist/server.js"]
    }
  }
}

Required: STRAVA_CLIENT_ID, STRAVA_CLIENT_SECRET (from a Strava API application with the authorization callback domain set to localhost). Optional: STRAVA_ACCESS_TOKEN/STRAVA_REFRESH_TOKEN (generated during setup), ROUTE_EXPORT_PATH (where exported route files are saved). Complete Strava's OAuth flow before restarting Claude.

Responses are returned as AI-friendly JSON via MCP rather than raw Strava API payloads, so an LLM can consume activity, segment, and route data directly. The documented setup sequence is exact and ordered - 1) install the server, 2) configure it in Claude, 3) complete the Strava authorization process, 4) restart Claude - and skipping or reordering these steps, or leaving the authorization callback domain unset to localhost, is called out as a likely source of setup failures. Token refresh is handled automatically once the initial OAuth flow completes, so the access and refresh tokens generated during setup don't need manual renewal.

Who it's for

Athletes and fitness enthusiasts who want Claude to pull up their Strava activities, stats, and route/segment data conversationally instead of opening the Strava app.

Source README

Strava MCP Server

Talk to your Strava data using AI.

Connect Claude to your Strava account and ask questions in plain English: "How far did I run this month?", "Analyze my last ride", or "Show me my fastest segments."

MseeP.ai Security Assessment Badge

Strava Server MCP server

What Can You Do With This?

Once connected, just talk to Claude like you're talking to a friend who has access to all your Strava data:

πŸƒ Track Your Progress

"How many kilometers did I run this month?"

"Compare my running stats from January to December"

"What's my longest ride ever?"

πŸ“Š Analyze Your Workouts

"Break down my last cycling workout - show me power, heart rate, and cadence"

"How did my heart rate zones look during yesterday's run?"

"What was my average pace for each lap in my interval training?"

πŸ—ΊοΈ Explore Routes & Segments

"What are the most popular cycling segments near Central Park?"

"Show me my starred segments"

"Export my Sunday morning route as a GPX file"

πŸ† Get Coaching Insights

"Analyze my training load this week"

"How does my current fitness compare to last month?"

"Give me a summary of my cycling performance this year"


Quick Start (3 Steps)

Step 1: Add to Claude Desktop

Open your Claude Desktop configuration file:

  • Mac: ~/Library/Application Support/Claude/claude_desktop_config.json
  • Windows: %APPDATA%\Claude\claude_desktop_config.json

Add this to the file:

{
  "mcpServers": {
    "strava": {
      "command": "npx",
      "args": ["-y", "@r-huijts/strava-mcp-server"]
    }
  }
}

Step 1 (alternative): Add to Claude Code

You can add this MCP server to claude code with the following command:

claude mcp add --transport stdio strava -- npx @r-huijts/strava-mcp-server

You can confirm successful instalation as follows:

% claude mcp list
Checking MCP server health...

strava: npx @r-huijts/strava-mcp-server - βœ“ Connected

Step 2: Restart Claude Desktop

Close and reopen Claude Desktop to load the new configuration.

Step 3: Connect Your Strava

Just say to Claude:

"Connect my Strava account"

A browser window will open. Enter your Strava API credentials, authorize the app, and you're done!

That's it! Start asking about your activities.


Connecting Your Strava Account

First Time Setup

When you say "Connect my Strava account", here's what happens:

  1. A browser window opens showing a setup page
  2. Enter your Strava API credentials (Client ID and Client Secret)
  3. Click "Continue to Strava" - you'll be redirected to Strava
  4. Authorize the app on Strava's website
  5. See the success message - you can close the browser
  6. Done! Claude confirms you're connected

Your credentials are saved locally at ~/.config/strava-mcp/config.json and persist across sessions.

Getting Your Strava API Credentials

You need to create a free Strava API application (one-time setup):

  1. Go to strava.com/settings/api
  2. Click "Create an App" (or view your existing app)
  3. Fill in the form:
    • Application Name: Anything you want (e.g., "My Claude Assistant")
    • Category: Choose any
    • Website: Can be anything (e.g., http://localhost)
    • Authorization Callback Domain: Must be localhost
  4. Copy your Client ID and Client Secret

That's it! You only need to do this once.

Managing Your Connection

  • Check status: "Am I connected to Strava?"
  • Reconnect: "Connect my Strava account" (use force: true to reconnect)
  • Disconnect: "Disconnect my Strava account"

Installation Options

Option A: Just Use It (Recommended)

No installation needed! The npx command in the Quick Start automatically downloads and runs the latest version.

Option B: Install Globally

If you prefer to install it once:

npm install -g @r-huijts/strava-mcp-server

Then update your Claude config to use:

{
  "mcpServers": {
    "strava": {
      "command": "strava-mcp-server"
    }
  }
}

Note: Even though the package name is @r-huijts/strava-mcp-server, the executable name remains strava-mcp-server for backward compatibility.

Option C: Build from Source

For developers who want to modify the code:

git clone https://github.com/r-huijts/strava-mcp.git
cd strava-mcp
npm install
npm run build

Then point Claude to your local build:

{
  "mcpServers": {
    "strava": {
      "command": "node",
      "args": ["/path/to/strava-mcp/dist/server.js"]
    }
  }
}

Example Conversations

Morning Check-in

You: What did I do for exercise this week?

Claude: Looking at your Strava activities this week, you completed:

  • Monday: 5.2km run (28:15)
  • Wednesday: 45km bike ride (1:32:00)
  • Saturday: 10km run (52:30)

Total: 60.2km across 3 activities. Nice consistency!

Workout Analysis

You: Analyze my last cycling workout in detail

Claude: Here's a breakdown of your ride "Evening Loop" from yesterday:

Overview

  • Distance: 42.3 km
  • Time: 1:28:45
  • Avg Speed: 28.6 km/h

Power Analysis

  • Average Power: 185W
  • Normalized Power: 198W
  • Max Power: 520W

Heart Rate

  • Average: 142 bpm
  • Max: 175 bpm
  • Time in Zone 4+: 23 minutes

Your power-to-heart-rate ratio suggests good aerobic efficiency...

Finding New Routes

You: Find me some challenging cycling climbs near Boulder, Colorado

Claude: Here are popular climbing segments near Boulder:

  1. Flagstaff Road - 5.2km, 6.8% avg grade
  2. Sunshine Canyon - 3.8km, 7.2% avg grade
  3. Left Hand Canyon - 8.1km, 4.5% avg grade

Want me to star any of these so you can find them easily later?


Available Tools

Account & Profile

What you can ask What it does
"Connect my Strava account" Links your Strava to Claude
"Check my Strava connection" Shows connection status
"Disconnect my Strava account" Unlinks Strava and clears stored tokens
"Get my Strava profile" Shows your profile info
"What shoes do I have?" Lists your shoes and usage distance
"What are my training zones?" Shows HR and power zones

Activities

What you can ask What it does
"Show my recent activities" Lists your latest workouts
"Get all my runs from January" Fetches activities with filters
"Analyze activity 12345" Detailed info about one activity
"Show the laps from my last run" Lap-by-lap breakdown
"Get heart rate data from my ride" Time-series workout data (optimized compact format)
"Show photos from my hike" Activity photos

Stats & Progress

What you can ask What it does
"What are my running stats?" YTD and all-time totals
"How far have I cycled this year?" Activity totals by type
"What's my longest ride?" Personal records

Segments

What you can ask What it does
"Show my starred segments" Your favorite segments
"Find segments near [location]" Discover popular segments
"Get details on segment 12345" Segment info and leaderboard
"Star this segment" Save to favorites
"Show my efforts on [segment]" Your attempts on a segment

Routes

What you can ask What it does
"List my saved routes" Your created routes
"Get details for my [route name]" Route info
"Export [route] as GPX" Download for GPS devices
"Export [route] as TCX" Download for GPS devices

Clubs

What you can ask What it does
"What clubs am I in?" Lists your Strava clubs

Workouts

What you can ask What it does
"Turn this workout into a Zwift file" Formats a workout plan into a structured file (currently Zwift .zwo)

Server

What you can ask What it does
"What version is the Strava server?" Shows the MCP server version and metadata

Troubleshooting

"Connect my Strava account" doesn't open a browser

  • Make sure Claude Desktop is running the MCP server (check for errors in Claude)
  • Try restarting Claude Desktop
  • Check that port 8111 isn't blocked by a firewall

"Authentication failed" or "Invalid token"

  • Your token may have expired - say "Connect my Strava account" to reconnect
  • Make sure your Strava API application is still active at strava.com/settings/api

"Missing credentials" error

  • You need to complete the OAuth flow - say "Connect my Strava account"
  • If you're using environment variables, make sure all 4 are set (see Developer section)

Claude doesn't see the Strava tools

  • Make sure your claude_desktop_config.json is valid JSON (no trailing commas!)
  • Restart Claude Desktop after making config changes
  • Check Claude's developer console for error messages

JSONRPC.ProtocolTransportError after package name change

If you're getting a JSONRPC error after updating to @r-huijts/strava-mcp-server:

  1. Clear npx cache: Run rm -rf ~/.npm/_npx in terminal
  2. Verify config uses @r-huijts/strava-mcp-server (not the old strava-mcp-server)
  3. Restart Claude Desktop completely (quit and reopen)
  4. Test manually: Run npx -y @r-huijts/strava-mcp-server - you should see "Starting Strava MCP Server v1.2.1..."

For Developers

Click to expand technical details

Environment Variables

Instead of using the browser-based auth, you can set environment variables:

Variable Description
STRAVA_CLIENT_ID Your Strava Application Client ID
STRAVA_CLIENT_SECRET Your Strava Application Client Secret
STRAVA_ACCESS_TOKEN OAuth access token
STRAVA_REFRESH_TOKEN OAuth refresh token
ROUTE_EXPORT_PATH Directory for GPX/TCX exports

Token Refresh

The server automatically refreshes expired tokens. New tokens are saved to both process.env and ~/.config/strava-mcp/config.json.

Config Priority

  1. Environment variables (highest)
  2. ~/.config/strava-mcp/config.json
  3. Local .env file

Building & Testing

npm install
npm run build
npm test

Activity Streams Optimization

The get-activity-streams tool uses a compact format by default, reducing payload size by ~70-80% while preserving all data:

  • Compact format (default): Raw arrays with metadata, ~70-80% smaller, ideal for LLM processing
  • Verbose format: Human-readable objects with formatted values (backward compatible)
  • Smart chunking: Large activities automatically split into ~50KB chunks
  • Optional downsampling: Can reduce very large datasets while preserving key features

The compact format includes comprehensive metadata (units, descriptions, statistics) so LLMs can understand the raw numeric data.

API Reference

The server implements the Model Context Protocol (MCP) and exposes 25 tools for Strava API v3. See the source code in src/tools/ for implementation details.

FAQ

Common questions

Discussion

Questions & comments Β· 0

Sign In Sign in to leave a comment.