Skip to main content
Glama

How to connect Strava to Claude Code, ChatGPT and any AI agent

Yes, Claude, ChatGPT and other AI agents can work in your Strava account. Strava's official Claude connector is read-only and only for Strava subscribers, and Strava has no official ChatGPT app. Cotlet.dev connects your own signed-in Strava account to any AI agent, so you can ask in plain words and the AI lists your activities and updates their gear for you.

What you can ask

List all my rides from the last 7 days with distance, moving time and elevation gain.

Set the gear on every run since October 1 to my Nike Pegasus 41.

Change the bike on Saturday's 82 km ride to my Canyon Endurace.

Show my 10 longest runs this month and which shoes I wore for each one.

How many km are on my Hoka Clifton 9? List the runs logged with them since August.

Related MCP server: Strava MCP Server

Connect in 2 minutes

Get your API key at https://cotlet.dev/app and export it as COTLET_API_KEY.

Claude Code

claude mcp add --transport http cotlet https://cotlet.dev/mcp/v1 --header "Authorization: Bearer $COTLET_API_KEY"

ChatGPT

Settings → Apps → add connector URL https://cotlet.dev/mcp/v1, then sign in with Cotlet.dev. No API key needed.

Cursor

Copy .cursor/mcp.json from this repo into your project:

{
  "mcpServers": {
    "cotlet": {
      "url": "https://cotlet.dev/mcp/v1",
      "headers": {
        "Authorization": "Bearer ${COTLET_API_KEY}"
      }
    }
  }
}

Run it locally (stdio)

This repo is a real MCP server. It speaks MCP over stdio, lists its Strava tools without any key, and forwards every tool call to the hosted Cotlet.dev server. The hosted URL above stays the recommended way. Use this when your client only starts local servers.

{"mcpServers":{"strava":{"command":"uvx","args":["strava-cotlet-mcp"],"env":{"COTLET_API_KEY":"your-key"}}}}

Or with Docker:

docker build -t strava-mcp .
docker run -i --rm -e COTLET_API_KEY strava-mcp

Tools: strava_guide, strava_status, strava_open, strava_read, strava_click, strava_type, strava_close. Without COTLET_API_KEY a tool call answers: "Set COTLET_API_KEY. Get a key at https://cotlet.dev/app".

Command line

Python 3.10+, nothing else to install. From PyPI:

pip install strava-cotlet-mcp        # or run it once: uvx strava-cotlet-mcp connect
strava-cotlet-mcp connect

Or straight from this repo:

export COTLET_API_KEY="your-key"
python3 strava_agent.py connect   # ready config for Claude Code, Cursor, ChatGPT
python3 strava_agent.py status    # is Strava ready in your Cotlet.dev account
python3 strava_agent.py guide     # Strava screens and sign-in link
python3 strava_agent.py read      # open Strava, read the page, close the browser
python3 strava_agent.py tools     # list Cotlet.dev tools
python3 strava_agent.py serve     # start the stdio MCP server (same as no argument)

Questions from Strava users

Does Strava have an API?

Strava has an official API and, since 1 June 2026, an official MCP connector (checked 27 Oct 2026). The MCP is read-only, works with Claude only and needs a paid Strava subscription. Cotlet.dev works in your own signed-in Strava account, so your AI agent can list activities and update gear with no API key and no code.

Is there an official Strava integration for ChatGPT?

Strava has no official ChatGPT app. It lists Claude and Gemini as its supported assistants. Add Cotlet.dev as an app in ChatGPT (Settings → Apps, connector URL https://cotlet.dev/mcp/v1), sign in to Strava once in your Cotlet.dev Cloud Browser, and ask ChatGPT in plain words. No code and no Strava API key.

Is there a Strava connector for Claude?

Yes. Strava has an official read-only Claude connector for subscribers. To also update activity gear, run the one-line claude mcp add command above in Claude Code. In the Claude app, add https://cotlet.dev/mcp/v1 as a custom connector. Claude then works in your own Strava account through Cotlet.dev.

Is there a Strava MCP server?

Strava offers an official, Anthropic-verified connector for Claude with read-only access for subscribers. Cotlet.dev is a hosted MCP server that works in Strava for you, 24/7, with Claude, ChatGPT, Cursor, Codex, n8n, Make and Zapier, including updating activity gear.

Find it on

Also works with


More: https://cotlet.dev/integrations/strava-api-mcp-server?utm_source=github&utm_medium=readme&utm_campaign=strava

Start free: https://cotlet.dev/signup?app=strava&utm_source=github&utm_medium=readme&utm_campaign=strava

Available Tools

7 tools
strava_clickBInspect

Click a numbered element on the current Strava page.

Inputs: element (required).

strava_closeAInspect

Close the Strava browser session when the work is done, so credits stop running.

Inputs: no named parameters.

strava_guideAInspect

How to work in Strava (list activities and update gear): its screens and the sign-in link. Call this first.

Inputs: task.

strava_openBInspect

Open a Strava page in your signed-in Cloud Browser (default https://www.strava.com) to work with list activities and update gear.

Inputs: url.

strava_readAInspect

Read the current Strava page: its text and numbered buttons, links and fields.

Inputs: element.

strava_statusAInspect

Check that your Cotlet.dev account and Strava are ready: account, credits and Strava connection.

Inputs: no named parameters.

strava_typeBInspect

Type text into a numbered field on the current Strava page.

Inputs: text (required), element (required).

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 7 tools

    Disambiguation5/5

    Each tool maps to a distinct step in a browser-automation lifecycle: guide, status check, open, read, click, type, close. There is no meaningful overlap between them, and the workflow ordering is explicit (guide first, close last).

    Naming Consistency5/5

    All seven tools use the same strava_ prefix followed by a short lowercase verb or noun (guide, status, open, read, click, type, close). The pattern is uniform and immediately readable as a single coherent family.

    Tool Count5/5

    Seven tools is a tight, well-scoped set for driving a signed-in browser session against one site. Every tool corresponds to a necessary primitive in that loop, with no redundant affordances.

    Completeness4/5

    The surface covers the essential interaction loop: orient (guide/status), navigate (open), inspect (read), act (click/type), and clean up (close). Minor gaps exist—no explicit scroll, back/navigation, wait-for-load, or file/upload helper—which an agent can usually work around by re-reading the page.

    Maintenance

    ActivityMaintained
    ResponsivenessOutdated data

    Related MCP Connectors

    Related MCP Servers