splitlog

Unofficial CLI and MCP server for Coros Training Hub — extract your activity data and health metrics from the command line. Designed for both humans and AI agents (Claude Desktop, Claude Code, scripts).

All commands produce rich terminal output for humans and structured JSON (--json) for agents and automation.

Not affiliated with, authorized, or endorsed by COROS. This tool accesses your own data using your own credentials.

Install

npm install -g @ibarker34/splitlog

Or run directly with npx:

npx -y @ibarker34/splitlog activities

Setup

Authenticate with your Coros account:

splitlog login

This stores your credentials in ~/.config/splitlog/config.json. Sessions are refreshed automatically — you only need to login once. EU accounts are auto-detected.

Usage

List activities

$ splitlog activities
                             Activities (191 total)
┌────────────┬────────────────┬───────────┬──────────┬──────────┬────────┬───────────┐
│ Date       │ Name           │ Type      │ Distance │ Duration │ Avg HR │ Elevation │
├────────────┼────────────────┼───────────┼──────────┼──────────┼────────┼───────────┤
│ 2026-03-04 │ Morning Run    │ Run       │ 6.58 km  │    41:29 │    174 │     +83m  │
│ 2026-03-03 │ Easy Run       │ Run       │ 3.80 km  │    29:12 │    151 │     +47m  │
└────────────┴────────────────┴───────────┴──────────┴──────────┴────────┴───────────┘

Options: --page, --size, --json

Activity detail

$ splitlog activity 475835095599579138
╭──────────────────────────── Morning Run ─────────────────────────────╮
│ Sport:          Run                                                   │
│ Distance:       6.58 km                                               │
│ Avg Pace:       6:17/km                                               │
│ Best Lap Pace:  4:44/km                                               │
│ Duration:       41:29                                                 │
│ Avg HR:         174 bpm                                               │
│ Aerobic Effect: 3.0                                                   │
│ Training Goal:  VO2 Max                                               │
╰───────────────────────────────────────────────────────────────────────╯

Includes laps, HR zones, and pace zones. The label ID comes from splitlog activities.

Health dashboard

$ splitlog health
 Running Level: 75.0
┌───────────┬───────┐
│ Category  │ Score │
├───────────┼───────┤
│ Endurance │  77.3 │
│ Threshold │  70.8 │
│ Speed     │  71.1 │
│ Sprint    │  70.7 │
└───────────┴───────┘
╭────────────────── Training Level ──────────────────╮
│ Status:        Optimal                              │
│ Load (ATI):    69                                   │
│ Fitness (CTI): 63                                   │
│ Trend:         109%                                 │
╰─────────────────────────────────────────────────────╯
            Race Predictor
┌───────────────┬─────────┬──────────┐
│ Distance      │    Time │ Avg Pace │
├───────────────┼─────────┼──────────┤
│ 5K            │   23:59 │  4:48/km │
│ 10K           │   50:00 │  5:00/km │
│ Semi Marathon │ 1:52:54 │  5:21/km │
│ Marathon      │ 4:00:17 │  5:42/km │
└───────────────┴─────────┴──────────┘

Also shows: resting HR, threshold HR/pace, recovery %, night HRV, weekly distance progress.

JSON output

All commands support --json for raw API data — useful for scripting or AI agents:

splitlog activities --json | jq '.data.dataList[0].name'
splitlog health --json | jq '.dashboard.data.summaryInfo.staminaLevel'

MCP Server (Claude Desktop)

splitlog includes a built-in MCP server, so Claude Desktop (or any MCP client) can access your training data directly.

Setup

1. Run splitlog login in your terminal first
2. Add to your Claude Desktop config (~/Library/Application Support/Claude/claude_desktop_config.json):

{
  "mcpServers": {
    "splitlog": {
      "command": "npx",
      "args": ["-y", "@ibarker34/splitlog", "mcp"]
    }
  }
}

3. Restart Claude Desktop — the tools will appear automatically

Available tools

Tool	Description
list_activities	List recent activities (paginated)
get_activity	Activity detail — full by default (laps, HR timeline, zones), or detailed=false for lightweight summary (multi-run analysis)
get_health	Health dashboard (running level, training status, race predictor, HRV)

How authentication works

When you run splitlog login:

1. Your password is hashed (MD5) locally before leaving your machine
2. The hash is sent to the login endpoint to obtain a session token
3. The token and password hash are saved in ~/.config/splitlog/config.json

Your plain-text password is never stored. When the session token expires, splitlog automatically re-authenticates using the stored hash — no manual re-login needed. You can inspect or delete the config file at any time.

Disclaimer

This is an unofficial tool, not affiliated with or endorsed by COROS. It accesses your own data using your own credentials. APIs may change at any time, which could break this tool. Use at your own risk.