io.github.hhopke/intervals-icu-mcp
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@io.github.hhopke/intervals-icu-mcp Show me my activities from the last 7 days."
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Intervals.icu MCP Server

A Model Context Protocol (MCP) server for Intervals.icu integration. Access your training data, wellness metrics, and performance analysis through Claude, ChatGPT, and other LLMs.
Originally based on eddmann/intervals-icu-mcp (MIT licensed). This project is an independent continuation with significant bug fixes and new features — see CHANGELOG.md for details.
Overview
67 tools spanning activities, activity analysis, activity messages, athlete profile, wellness, events/calendar, performance curves, workout library, gear, sport settings, and custom items — plus 4 MCP Resources (athlete profile, workout syntax, event categories, custom item schemas) and 7 MCP Prompts (training analysis, recovery check, weekly planning, and more). See Available Tools for the per-category breakdown.
Related MCP server: Intervals.icu MCP Server
Quick Start
Or for Claude Desktop, in 30 seconds:
Get your API key and athlete ID
Add this to your Claude Desktop config:
{
"mcpServers": {
"intervals-icu": {
"command": "uvx",
"args": ["intervals-icu-mcp"],
"env": {
"INTERVALS_ICU_API_KEY": "your-api-key-here",
"INTERVALS_ICU_ATHLETE_ID": "i123456"
}
}
}
}Restart Claude and ask "Show me my activities from the last 7 days."
Prefer Claude Code, Cursor, or ChatGPT? See Client Configuration. Want to run from source or with Docker? See Installation & Setup.
Prerequisites
Install uv — it handles Python, dependencies, and execution in one tool. brew install uv on macOS/Linux, or powershell -c "irm https://astral.sh/uv/install.ps1 | iex" on Windows. From there, uvx fetches Python and the package automatically. Docker is also supported as an alternative.
Intervals.icu API Key Setup
Before installation, obtain your Intervals.icu API key:
Go to https://intervals.icu/settings → Developer → Create API Key.
Copy the key, and note your Athlete ID from your profile URL (format:
i123456).
Installation & Setup
Nothing to install separately if you use the recommended setup. uvx (which ships with uv) automatically downloads and caches the intervals-icu-mcp package the first time your MCP client launches it — just paste the config snippet from Client Configuration into your client and you're done.
git clone https://github.com/hhopke/intervals-icu-mcp.git
cd intervals-icu-mcp
uv sync
uv run intervals-icu-mcp-auth # interactive credential setup; or create .env manually:
# INTERVALS_ICU_API_KEY=your_api_key_here
# INTERVALS_ICU_ATHLETE_ID=i123456Then point your MCP client at this checkout — see the From source snippet inside each client below.
docker build -t intervals-icu-mcp .
# Interactive credential setup (creates intervals-icu-mcp.env in the current directory):
touch intervals-icu-mcp.env # pre-create the file so Docker mounts it as a file, not a dir
docker run -it --rm \
-v "$(pwd)/intervals-icu-mcp.env:/app/.env" \
--entrypoint= intervals-icu-mcp:latest \
python -m intervals_icu_mcp.scripts.setup_authOr create intervals-icu-mcp.env manually (same format as the .env above).
Then point your MCP client at the Docker image — see the Docker snippet inside each client below.
Client Configuration
The server speaks MCP over stdio and works with any compliant client. Click a client to expand. If you followed Quick Start (uvx), use the first config block; if you used the source or Docker alternative above, use the matching variant inside the same collapsible.
Add to your configuration file:
macOS:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"intervals-icu": {
"command": "uvx",
"args": ["intervals-icu-mcp"],
"env": {
"INTERVALS_ICU_API_KEY": "your-api-key-here",
"INTERVALS_ICU_ATHLETE_ID": "i123456"
}
}
}
}From source (requires git clone + uv sync + uv run intervals-icu-mcp-auth):
{
"mcpServers": {
"intervals-icu": {
"command": "uv",
"args": ["run", "--directory", "/ABSOLUTE/PATH/TO/intervals-icu-mcp", "intervals-icu-mcp"]
}
}
}Docker:
{
"mcpServers": {
"intervals-icu": {
"command": "docker",
"args": ["run", "-i", "--rm", "-v", "/ABSOLUTE/PATH/TO/intervals-icu-mcp.env:/app/.env", "intervals-icu-mcp:latest"]
}
}
}Register the server as a user-scoped MCP server:
claude mcp add intervals-icu --scope user \
--env INTERVALS_ICU_API_KEY=your-key \
--env INTERVALS_ICU_ATHLETE_ID=i123456 \
-- uvx intervals-icu-mcpThen in any Claude Code session, run /mcp to confirm intervals-icu is connected.
Add to ~/.cursor/mcp.json (or the project-local .cursor/mcp.json):
{
"mcpServers": {
"intervals-icu": {
"command": "uvx",
"args": ["intervals-icu-mcp"],
"env": {
"INTERVALS_ICU_API_KEY": "your-api-key-here",
"INTERVALS_ICU_ATHLETE_ID": "i123456"
}
}
}
}Restart Cursor and open Settings → MCP to verify the server is listed.
ChatGPT's custom MCP connector flow requires running the server over HTTP and exposing it via a tunnel, then registering the URL in ChatGPT's Developer Mode settings. See docs/chatgpt-connector.md for the full walkthrough, plan-tier requirements, and security notes.
Usage
Ask Claude to interact with your Intervals.icu data in natural language. A few starter prompts:
"Show me my activities from the last 30 days"
"Am I overtraining? Check my CTL, ATL, and TSB"
"How's my recovery this week? Show HRV and sleep trends"
"Create a sweet spot cycling workout for tomorrow"
"What's my 20-minute power and FTP?"For the full catalogue of example prompts by category, see docs/examples.md.
Available Tools
67 tools, 4 resources, and 7 prompt templates. One-line summary below — full reference in docs/tools.md.
Category | Tools | Summary |
12 | Query, search, update, delete, download activities | |
8 | Streams, intervals, best efforts, histograms | |
2 | Read and post notes/comments/coach feedback on activities | |
3 | Profile, CTL/ATL/TSB analysis, and fitness chart time-series | |
3 | HRV, sleep, recovery metrics | |
11 | Planned workouts, races, notes, ATP periodization (bulk ops supported) | |
3 | Power, HR, and pace curves with zones | |
7 | Browse and create folders and training plans; create (incl. bulk), update, and delete library workouts | |
6 | Track equipment and maintenance reminders | |
5 | FTP, FTHR, pace thresholds, and zones | |
5 | User customizations: custom charts, fields, zones, dashboard panels |
Delete Safety Mode
Destructive tools are gated by the optional INTERVALS_ICU_DELETE_MODE env var (safe / full / none, default safe) — a server-side gate outside the model's reach, so unregistered tools can't be invoked. See docs/tools.md for the full mode table, response envelope, and TZ-buffer rationale.
Remote Deployment (HTTP / SSE)
The server runs over stdio by default — the right transport for local clients like Claude Desktop, Claude Code, and Cursor. HTTP and SSE transports are available for remote or hosted use.
⚠️ MCP has no built-in authentication — never expose an HTTP-mode server to an untrusted network without a tunnel (Tailscale, Cloudflare Tunnel) or an authenticating reverse proxy.
See docs/remote-deployment.md for transport flags and the full security model.
Documentation
Example prompts — full catalogue of natural-language prompts by category
Tool reference — complete tool, resource, and prompt inventory
Architecture overview — how the server, middleware, client, and tools fit together
Remote deployment (HTTP/SSE) — transports, flags, and the security model for hosted/remote setups
Testing guide — conventions for pytest + respx, fixtures, and running the suite
Changelog — release history
Adding a new tool — step-by-step workflow for contributors
Feedback
How are you using this? Which tools you lean on, what your prompts look like, where it gets in your way — that shapes the roadmap more than my own guesses do.
Show and tell · Q&A · Ideas · General — or open an issue for a reproducible bug.
Contributing
Contributions are very welcome, and none is too small — a typo, a clearer parameter description, an extra test, a whole new tool. No Python or MCP expertise assumed. See CONTRIBUTING.md; in short, run make can-release before opening a PR. Good first issues are a gentle place to start.
Sponsor
Free under the MIT License, maintained in my spare time. If it's useful to you, GitHub Sponsors supports continued development — entirely optional, and feedback or a PR helps just as much.
License
MIT License - see the LICENSE file for details.
Disclaimer
This project is not affiliated with, endorsed by, or sponsored by Intervals.icu. All product names, logos, and brands are property of their respective owners.
Available Tools
64 toolsicu_add_activity_messageA
POST a new note or comment on a specific activity (singular = WRITE).
Use when the user wants to leave a note on one of their activities: "add a note to this ride that I felt strong", "comment on yesterday's run", "leave a training note saying...". Attributed to the authenticated user. To READ existing notes use icu_get_activity_messages (plural).
Inputs: content (required), activity_id (required). Returns: result
icu_apply_sport_settingsADestructive
Recompute training load, zones, and derived metrics on HISTORICAL activities using the current sport settings.
Different from update_sport_settings (which just stores new values). Use after changing FTP/FTHR/pace to backfill chart math.
Inputs: sport_id (required), athlete_id. Returns: result
icu_apply_training_planADestructive
Schedule an entire training plan (workout-library folder) onto the athlete's calendar starting on a chosen date.
Use after icu_get_workout_library to find a plan's folder_id. Different from icu_create_event / icu_bulk_create_events (which build new events) and from icu_duplicate_events (which copies existing).
Inputs: folder_id (required), start_date_local (required), athlete_id, extra_workouts_json. Returns: result
icu_bulk_create_eventsA
Create MANY new calendar events in a single batch call (more efficient than looping create_event).
Accepts a JSON array of event objects, each shaped like an icu_create_event payload. For copying existing events forward use icu_duplicate_events instead — that reuses payloads rather than taking new fields. See intervals-icu://event-categories and intervals-icu://workout-syntax for the referenced enums and DSL.
Inputs: events (required), athlete_id. Returns: result
icu_bulk_create_manual_activitiesADestructive
Batch-create manual activities (no device upload) with UPSERT on external_id.
Existing activities with a matching external_id (set by the same OAuth app) are updated; activities without an external_id are always created new.
Inputs: activities_json (required), athlete_id. Returns: result
icu_bulk_create_workoutsA
Save MANY new reusable workouts into LIBRARY folders or plans in one call — e.g. filling out a training plan.
Each item is shaped like an icu_create_workout call. Nothing lands on the calendar: for dated calendar workouts use icu_bulk_create_events, and to schedule a finished plan use icu_apply_training_plan.
Inputs: workouts (required), athlete_id. Returns: result
icu_bulk_delete_eventsADestructiveIdempotent
Delete MANY calendar events in a single batch call. Destructive — cannot be undone.
In safe delete mode (default), the call partitions the input list
into deleted (future) and skipped (past or undated) and returns
both. INTERVALS_ICU_DELETE_MODE=full disables the partition.
Inputs: event_ids (required), athlete_id. Returns: result
icu_create_custom_itemA
Create a custom addition to the user's Intervals.icu account.
Use when the user says things like: "add a custom field for RPE", "create
a custom power zone set", "add a chart for monthly distance". Match the
user's intent to the right item_type (see that param's description). For
the content schema per item_type, read intervals-icu://custom-item-schemas.
Inputs: name (required), item_type (required), content, athlete_id, +2 more. Returns: result
icu_create_eventA
Create ONE new calendar event from scratch.
For two or more events in a single call, prefer icu_bulk_create_events over a loop. For copying existing events forward in time (repeating a workout for N weeks), use icu_duplicate_events — that tool reuses an existing event's payload instead of taking new fields.
For category guidance and the training_availability enum, read the
intervals-icu://event-categories resource. For structured WORKOUT events,
put workout-syntax text in description — see intervals-icu://workout-syntax.
Inputs: name (required), category (required), start_date (required), color, +11 more. Returns: result
icu_create_gearB
Create a new gear item for tracking equipment usage and maintenance (bikes, shoes, trainers, etc.).
Inputs: name (required), gear_type (required), active, athlete_id. Returns: result
icu_create_gear_reminderA
Create a maintenance reminder for a gear item, triggered by distance, time, or both.
Inputs: text (required), gear_id (required), athlete_id, time_alert, +1 more. Returns: result
icu_create_sport_settingsB
Create a per-sport threshold record with outdoor/indoor FTP, FTHR, pace, or swim settings.
Inputs: sport_type (required), ftp, fthr, athlete_id, +3 more. Returns: result
icu_create_workoutA
Save ONE new reusable workout into a LIBRARY folder or training plan — it does not go on the calendar.
To schedule a workout on a date use icu_create_event. For two or more library workouts use icu_bulk_create_workouts; to change a saved one use icu_update_workout. Make a new folder or plan with icu_create_workout_folder, and put a plan's workouts on the calendar with icu_apply_training_plan.
Inputs: name (required), folder_id (required), day, tags, +9 more. Returns: result
icu_create_workout_folderA
Create a new LIBRARY folder or training PLAN to hold workouts — not a calendar item.
Pass the returned id as folder_id to icu_create_workout or icu_bulk_create_workouts. To browse existing folders use icu_get_workout_library.
Inputs: name (required), athlete_id, description, folder_type. Returns: result
icu_delete_eventADestructiveIdempotent
Permanently delete ONE calendar event by ID. Destructive — cannot be undone.
In safe delete mode (default), past events are refused and reported
in the skipped envelope with a hint about INTERVALS_ICU_DELETE_MODE=full.
Returns a deleted / skipped envelope either way.
Inputs: event_id (required), athlete_id. Returns: result
icu_delete_gearADestructiveIdempotent
Permanently delete a gear item and its maintenance reminders. Activities that used this gear are not affected.
Inputs: gear_id (required), athlete_id. Returns: result
icu_delete_workoutADestructiveIdempotent
Permanently delete ONE workout from the LIBRARY. Destructive — cannot be undone.
Removes the library workout only; calendar events are separate records deleted with icu_delete_event. Confirm with the user before calling if unsure.
Inputs: workout_id (required), athlete_id. Returns: result
icu_download_activity_fileARead-onlyIdempotent
Download the original activity file.
Downloads the ORIGINAL uploaded file (FIT, TCX, or GPX — whatever the
device produced). Different from download_fit_file (forces FIT conversion)
and download_gpx_file (forces GPX conversion). If output_path is set
the file is saved there; otherwise the response embeds base64 content.
Inputs: activity_id (required), output_path. Returns: result
icu_download_fit_fileARead-onlyIdempotent
Download activity converted to FIT format (Garmin / most training platforms). Different from download_activity_file (original upload format) and download_gpx_file (GPX format).
Inputs: activity_id (required), output_path. Returns: result
icu_download_gpx_fileARead-onlyIdempotent
Download activity converted to GPX format (GPS devices, mapping software). Different from download_activity_file (original upload format) and download_fit_file (FIT format).
Inputs: activity_id (required), output_path. Returns: result
icu_duplicate_eventsA
COPY existing events forward in time by N weeks.
Use when the user says "repeat this workout for the next 4 weeks", "duplicate Monday's run on the next 3 Mondays". Reuses the existing events' payloads — different from icu_create_event / icu_bulk_create_events, which both build NEW events from scratch.
Inputs: event_ids (required), athlete_id, num_copies, weeks_between. Returns: result
icu_get_activities_aroundARead-onlyIdempotent
Fetch the activities chronologically before and after a reference activity (N each side).
Use for "what did I do around this race?", training-context queries, progression comparisons.
Inputs: activity_id (required), count, athlete_id. Returns: result
icu_get_activities_by_dateARead-onlyIdempotent
List activities within an EXPLICIT date window (oldest..newest) — LIGHT summary per item.
Unlike icu_get_recent_activities (anchored to today and capped at 100), this
targets an arbitrary historical window and is bounded only by limit. Use for
"all my runs from June to November 2025", reconstructing training history, or
finding the oldest activity in a period. Dates are YYYY-MM-DD. Results are
newest-first; if a window holds more than limit items, the oldest are
dropped first, so widen limit (or narrow the window) to reach the very
oldest.
Inputs: oldest (required), limit, newest, athlete_id. Returns: result
icu_get_activity_detailsARead-onlyIdempotent
Fetch the headline SUMMARY of one activity — name, sport, date, distance, duration, training load, weather, plus all top-level metrics in a single JSON blob.
Use for "how was my ride?", "tell me about activity X". For lap-by-lap or per-interval breakdown use icu_get_activity_intervals; for second-by-second time-series use icu_get_activity_streams.
Inputs: activity_id (required). Returns: result
icu_get_activity_intervalsARead-onlyIdempotent
Fetch the per-LAP / per-interval breakdown of one activity — each segment with its target, actual power/HR/pace, and type (warm-up / work / rest / cool-down).
Use for workout-compliance analysis, lap-by-lap review, "did I hit my intervals?". For headline summary metrics use get_activity_details; for raw second-by-second data use get_activity_streams.
Inputs: activity_id (required). Returns: result
icu_get_activity_messagesARead-onlyIdempotent
Read the notes and comments attached to a specific activity (plural = READ).
Use when the user asks: "what did my coach say about that ride?", "show me the comments on yesterday's run", "any feedback on this workout?". Returns every message in chronological order with author, content, timestamp, and seen-flag. To POST a new message use icu_add_activity_message (singular = WRITE).
Inputs: activity_id (required). Returns: result
icu_get_activity_streamsARead-onlyIdempotent
Fetch RAW per-sample time-series of one activity — second-by-second arrays for power, HR, cadence, speed, altitude, GPS, temperature, grade, etc.
Heavy payload. Use only when you need the underlying signal for visualization or custom analysis. Most "how was my ride?" questions are better answered by get_activity_details (summary metrics) or get_activity_intervals (per-lap breakdown).
Stream-type filter accepts any of: watts, heartrate, cadence, velocity_smooth, altitude, distance, time, latlng, temp, moving, grade_smooth, raw_heartrate, fixed_heartrate.
heartrate is CORRECTED data: Intervals.icu replaces readings above the
athlete's configured max HR with an interpolated line at import, so it can
never exceed that setting. raw_heartrate is the uncorrected trace and is
the only way to see a genuine new max that was clipped away — the platform
never raises the setting on its own. Request it explicitly; it is not in
the default set.
Inputs: activity_id (required), streams. Returns: result
icu_get_annual_training_planARead-onlyIdempotent
Read the Annual Training Plan (ATP) on the calendar — weekly load targets (TSS), Base/Build/Peak phases, and ATP week notes.
Use when the user asks: "what's my weekly TSS target?", "what training phase am I in?", "show my periodization plan", "when are my recovery weeks?", "what's the note on this training week?". Defaults to a 365-day forward window so the full ATP is returned; pass a smaller days_ahead/days_back when the user asks about a specific month or week. NOT individual planned workouts (icu_get_upcoming_workouts), NOT all calendar entries (icu_get_calendar_events), NOT workout-library plan folders (icu_get_workout_library / icu_apply_training_plan).
Inputs: days_back, athlete_id, days_ahead. Returns: result
icu_get_athlete_profileARead-onlyIdempotent
Get an athlete's profile — sport settings (outdoor/indoor FTP, FTHR, pace) and current CTL/ATL/TSB.
Defaults to the authenticated athlete; coaches can pass athlete_id to read one of their managed athletes instead.
Inputs: athlete_id. Returns: result
icu_get_best_effortsARead-onlyIdempotent
Find the top-N peak efforts WITHIN a single activity for a given stream + target duration or distance.
Different from get_*_curves (which span many activities). Useful for "what was my best 20-min power on this ride?" or "show my top 5k splits in this run." Requires at least one of duration or distance.
Inputs: activity_id (required), count, stream, distance, +1 more. Returns: result
icu_get_calendar_eventsARead-onlyIdempotent
Fetch ALL calendar entries in a date window — workouts, notes, races, goals, life-event blocks.
Use for "what's on my calendar?", "show this week", broad calendar queries. For just the planned WORKOUT entries (filtered) use icu_get_upcoming_workouts. For one specific event by ID use get_event.
Inputs: days_back, athlete_id, days_ahead. Returns: result
icu_get_custom_itemARead-onlyIdempotent
Fetch the full configuration of ONE custom addition by ID.
Use AFTER icu_get_custom_items has returned an ID when the user wants
to inspect a specific chart/field/zone/panel — the content field
here is the same as in the list, just focused on one item.
Inputs: item_id (required), athlete_id. Returns: result
icu_get_custom_itemsARead-onlyIdempotent
List the user's custom additions to their Intervals.icu account.
Use this when the user asks about THEIR OWN customizations: "show my custom charts", "list my custom fields", "what custom zones do I have", "what's on my dashboard", "do I have any custom activity panels".
Returns every custom item across all types in one call: custom charts
(FITNESS_CHART, TRACE_CHART, ACTIVITY_CHART, ACTIVITY_HISTOGRAM,
ACTIVITY_HEATMAP, ACTIVITY_MAP, ACTIVITY_PANEL, FITNESS_TABLE), custom
data fields (INPUT_FIELD on wellness, ACTIVITY_FIELD on activities,
INTERVAL_FIELD on intervals), custom ACTIVITY_STREAM definitions, and
custom ZONES configurations. Each item has a type field so you can
filter client-side if the user asked about a specific kind.
Do NOT use this for built-in zones, built-in fields, or athlete profile data — those have dedicated tools (icu_get_sport_settings, etc.).
Inputs: athlete_id. Returns: result
icu_get_eventARead-onlyIdempotent
Fetch ONE specific calendar event by ID — full details including description, workout structure, and metrics.
Inputs: event_id (required), athlete_id. Returns: result
icu_get_fitness_chartARead-onlyIdempotent
Fetch the Performance Management Chart time-series — daily CTL, ATL, and TSB.
Returns past history plus future projections from planned calendar workouts. Use for: "show my fitness chart", "CTL trend last 90 days", "projected form at end of my block", "what will my TSB be in 3 weeks". NOT for: today's training recommendations (icu_get_fitness_summary), HRV/sleep/ recovery trends (icu_get_wellness_data), or one day's full wellness record (icu_get_wellness_for_date).
Inputs: days_back (required), days_ahead (required), athlete_id. Returns: result
icu_get_fitness_summaryARead-onlyIdempotent
Get an athlete's current fitness / fatigue / form snapshot — CTL, ATL, TSB, ramp rate, with interpretation and training recommendations.
Use for "how's my form?", "am I overtrained?", training-status checks. Defaults to the authenticated athlete; coaches can pass athlete_id to read one of their managed athletes instead.
Inputs: athlete_id. Returns: result
icu_get_gap_histogramARead-onlyIdempotent
Time-in-zone DISTRIBUTION of grade-adjusted pace (GAP) within a single activity — elevation-normalized.
Use for trail running where raw pace is misleading. For raw (non- elevation-normalized) pace use icu_get_pace_histogram.
Inputs: activity_id (required). Returns: result
icu_get_gear_listBRead-onlyIdempotent
List all gear items with usage stats (distance, time, activity count) and maintenance reminders.
Inputs: athlete_id. Returns: result
icu_get_hr_curvesARead-onlyIdempotent
Fetch the HR-vs-duration curve — best (highest) sustained HR across durations from 5s up to 1h, aggregated over the chosen window.
Use for cardiovascular-fitness trends. For the athlete's HR zones and max HR, call icu_get_sport_settings — this tool does not return zone bands, and its peak_hr_bpm is the top of this curve, which Intervals.icu has already clipped to the configured max HR at import. For time-in-zone distribution within a single activity, use get_hr_histogram instead.
Inputs: days_back, athlete_id, sport_type, time_period. Returns: result
icu_get_hr_histogramARead-onlyIdempotent
Time-in-zone DISTRIBUTION of heart rate within a single activity (histogram buckets, time per bucket).
Different from icu_get_hr_curves (best efforts across many activities). Use for cardiovascular load breakdown and HR-zone time-in-zone.
Inputs: activity_id (required). Returns: result
icu_get_pace_curvesARead-onlyIdempotent
Fetch the pace-vs-duration curve — best (fastest) sustained pace across durations from 5s up to 1h, aggregated over the chosen window.
Use for run/swim fitness trends and race-time predictions. Pass
use_gap=True to normalize for hills via Grade-Adjusted Pace. For
time-in-zone distribution within a single activity, use
get_pace_histogram (or get_gap_histogram) instead.
Inputs: use_gap, days_back, athlete_id, sport_type, +1 more. Returns: result
icu_get_pace_histogramARead-onlyIdempotent
Time-in-zone DISTRIBUTION of pace within a single running activity (histogram buckets, time per bucket).
Different from icu_get_pace_curves (best efforts across many activities). Use for pace-distribution / consistency analysis. For elevation- normalized pace use icu_get_gap_histogram.
Inputs: activity_id (required). Returns: result
icu_get_power_curvesARead-onlyIdempotent
Fetch the power-vs-duration curve — best (highest) sustained watts across durations from 5s up to 1h, aggregated over the chosen window.
Use for FTP estimation, peak-power tracking, strengths/weaknesses across duration profiles. For time-in-zone distribution within a single activity, use icu_get_power_histogram instead.
Inputs: days_back, athlete_id, sport_type, time_period. Returns: result
icu_get_power_histogramARead-onlyIdempotent
Time-in-zone DISTRIBUTION of power within a single activity (histogram buckets, time per bucket).
Different from icu_get_power_curves (best efforts across many activities). Use for "how was my workout intensity distributed?", training-zone breakdown.
Inputs: activity_id (required). Returns: result
icu_get_recent_activitiesARead-onlyIdempotent
List the athlete's most recent activities (default last 30 days) — LIGHT summary per item (distance, duration, power, HR, training load).
Use for "what have I done recently?", "show last week's rides". For one specific activity by ID use icu_get_activity_details; to search by name/tag use icu_search_activities.
Inputs: limit, days_back, athlete_id. Returns: result
icu_get_sport_settingsARead-onlyIdempotent
Get per-sport thresholds and the athlete's configured power/HR/pace zones.
Returns outdoor/indoor FTP, FTHR, max HR, running pace and swim threshold, plus the zone sets configured in Intervals.icu (HR zones in bpm, power zones as %FTP, pace zones as % of threshold pace) with their names.
This is the ONLY source of the athlete's real zones. Intervals.icu derives them from the threshold and stamps them into every activity at import, so time-in-zone, HRSS and TSS are all computed from these — reasoning about zones from any other number puts the answer at odds with the athlete's own charts. Zones are not derived from curve data.
Inputs: athlete_id. Returns: result
icu_get_upcoming_workoutsARead-onlyIdempotent
Fetch the planned WORKOUT entries dated on the upcoming CALENDAR (filters out notes, races, goals) — these are calendar EVENTS, not workout-library templates.
Use for "what's my next workout?", "what training is planned". Each id is
a calendar event ID: pass it to icu_get_event / icu_update_event /
icu_delete_event, never to the icu_*_workout library tools. For every
calendar entry type use icu_get_calendar_events; for reusable templates
stored in the library use icu_get_workouts_in_folder.
Inputs: limit, athlete_id. Returns: result
icu_get_wellness_dataARead-onlyIdempotent
Fetch wellness records over a RANGE of recent days (default last 7).
Use for trends, weekly summaries, "how has my sleep been this week?", recovery curves. For PMC/fitness chart CTL/ATL/TSB series use icu_get_fitness_chart. For a single specific date use icu_get_wellness_for_date instead — this tool always returns a multi-day list.
Inputs: days_back, athlete_id. Returns: result
icu_get_wellness_for_dateARead-onlyIdempotent
Fetch the wellness record for ONE specific date.
Use when the user names a date — "show my HRV for Monday", "wellness on 2026-03-15", "how did I sleep last Thursday?". For ranges, weeks, or trends use icu_get_wellness_data.
Inputs: date (required), athlete_id. Returns: result
icu_get_workout_libraryARead-onlyIdempotent
List all workout-library folders + training plans the athlete has access to (personal, shared, and followed plans).
Each folder ID can be passed to icu_get_workouts_in_folder to see its contents, or to icu_apply_training_plan to schedule it onto the calendar.
Inputs: athlete_id. Returns: result
icu_get_workouts_in_folderBRead-onlyIdempotent
List the workouts stored in one specific library folder or training plan — name, type, structure, training load, intensity factor.
Inputs: folder_id (required), athlete_id. Returns: result
icu_list_athletesARead-onlyIdempotent
List the athletes this account can access — the caller plus anyone they follow or coach.
Call this FIRST whenever a request concerns someone other than the default athlete, to resolve a name to the athlete_id that other tools take. Returns each athlete's access level, so you know before calling whether a write will be permitted.
Inputs: no named parameters. Returns: result
icu_search_activitiesARead-onlyIdempotent
Search activities by name or tag, returning a LIGHT result list — id, name, type, date, distance, time only.
Use this first for "find my X" queries. Only escalate to search_activities_full when you specifically need power, HR, training load, or intensity-factor data on the matches (heavier payload).
Inputs: query (required), limit, athlete_id. Returns: result
icu_search_activities_fullARead-onlyIdempotent
Search activities by name or tag, returning FULL Activity objects with power, HR, training load, intensity factor, normalized power, weather — every metric per result.
Heavy payload. Use only when the lighter search_activities won't tell you what you need (e.g. "find my threshold rides with NP above 250W").
Inputs: query (required), limit, athlete_id. Returns: result
icu_search_intervalsARead-onlyIdempotent
Search intervals ACROSS ALL the athlete's activities (cross-activity).
Different from icu_get_activity_intervals (single-activity). Use to track
progress on a workout type ("all my threshold intervals over the last
year") or find comparable historical sessions. Returns the activities
containing matching intervals, each with an interval_summary like
"2x 8m 162w" (reps x duration at target).
Inputs: limit, athlete_id, max_duration, min_duration, +3 more. Returns: result
icu_update_activityAIdempotent
Update an existing activity's metadata (name, type, trainer flag, RPE, feel, etc.).
Only fields you pass are sent; everything else stays unchanged.
Inputs: activity_id (required), feel, name, commute, +4 more. Returns: result
icu_update_activity_streamsADestructive
Upload raw time-series streams (power, HR, cadence, etc.) onto an existing activity. Destructive — overwrites existing stream data.
Accepts JSON array or CSV. Different from get_activity_streams (READ).
Inputs: activity_id (required), payload_string (required), format. Returns: result
icu_update_custom_itemAIdempotent
Modify an existing custom chart/field/zones/panel.
Use when the user wants to change one of their existing customizations:
"rename my custom field", "make this chart public". Usually need
icu_get_custom_items first to find the right item_id. Only fields you
pass are sent — others are left unchanged. For content schema, see
intervals-icu://custom-item-schemas.
Inputs: item_id (required), name, content, item_type, +3 more. Returns: result
icu_update_eventAIdempotent
Update an existing calendar event.
Only fields you pass are sent — other fields remain unchanged. For category
and training_availability semantics, see intervals-icu://event-categories.
For WORKOUT description syntax, see intervals-icu://workout-syntax.
Inputs: event_id (required), name, color, end_date, +11 more. Returns: result
icu_update_gearAIdempotent
Update an existing gear item. Only fields you pass are sent.
Inputs: gear_id (required), name, active, gear_type, +1 more. Returns: result
icu_update_gear_reminderAIdempotent
Update an existing gear maintenance reminder. Only fields you pass are sent.
Inputs: gear_id (required), reminder_id (required), text, athlete_id, +2 more. Returns: result
icu_update_sport_settingsAIdempotent
Update an existing per-sport threshold record (outdoor/indoor FTP, FTHR, pace, swim).
Inputs: sport_id (required), ftp, fthr, athlete_id, +4 more. Returns: result
icu_update_wellnessAIdempotent
Upsert wellness data for ONE specific date — creates the record if missing, otherwise updates the fields you pass.
Only provided fields are sent. Subjective scales (fatigue, soreness,
stress, mood, motivation, injury) are 1-5. Pass locked=True to stop
device sync from overwriting manual entries. Writes to the authenticated
athlete unless athlete_id names a managed athlete.
Inputs: date (required), hrv, mood, spo2, +28 more. Returns: result
icu_update_workoutAIdempotent
Change fields on an EXISTING library workout — only the fields you pass are sent.
To save a new workout use icu_create_workout. This edits the library workout only; calendar events are separate records changed with icu_update_event.
Inputs: workout_id (required), day, name, tags, +10 more. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 64 tools
Each tool has a clearly distinct purpose, and the descriptions explicitly differentiate them from similar ones (e.g., histograms vs curves, single vs bulk operations). The thorough 'Different from X' notes eliminate ambiguity, making misselection unlikely despite the large number.
All tools follow a uniform `icu_verb_noun` pattern (e.g., icu_get_power_curves, icu_create_event, icu_bulk_delete_events). The prefix is consistent, verbs are clear and standard (get/create/update/delete), and no mixed conventions are present.
64 tools is extreme for an MCP server, far exceeding the 25+ threshold for 'too many'. Even for a broad training platform, this overwhelms an agent's tool selection and adds significant token overhead. Many operations could be combined or parameterized (e.g., single vs bulk variants).
The tool set covers the major domains—activities, workouts, calendar, gear, wellness, fitness, settings, and custom items—with CRUD-like operations. Minor gaps exist (no delete for custom items or sport settings, no single activity creation, no file upload), but these are edge cases that agents can work around.
Related MCP Connectors
Connect Intervals.icu to Claude and ChatGPT: read training, plan workouts. Training never stored.
Intervals.icu MCP server for ChatGPT, Claude and any AI agent: list activities, get activity with intervals, list wellness, list calendar events, create planned workout, update planned workout, delete calendar event and log daily wellness in your own Intervals.icu account. Cotlet works in the account through your Cloud Browser, a real Chrome where you sign in to Intervals.icu yourself; Cotlet does not store your password and you need no Intervals.icu API key. About the official Intervals.icu API: Yes, Intervals.icu has a public REST API (checked 6 Nov 2026). You generate a personal API key under Developer Settings for your own data, and apps for many users use OAuth. Setup and FAQ: https://cotlet.dev/integrations/intervals-icu-api-mcp-server
tri-train is a training platform for triathletes that syncs activities from Strava and Garmin and exposes them to Claude through a remote MCP server. Ask Claude how your last block went, why your run pace dropped, or whether you are fresh enough for Sunday's race, and it answers from your real data: activities with laps and grade-adjusted pace, weekly load and the performance management chart (CTL/ATL/TSB), sleep, HRV and resting heart rate, and your race calendar with predictions. Claude can also act: create and adjust planned workouts with structured intervals, push them to your Garmin watch, add races and write notes. Every change is journaled and can be undone. Read and write access are separate OAuth scopes, and the server enforces them on every call. Built and hosted in Switzerland by a triathlete; data stays on our own servers (Swiss revDSG / GDPR).
Adaptive running coach MCP server — training data, plans, and recovery for AI assistants.
Related MCP Servers
- AlicenseAqualityDmaintenanceEnables interaction with Intervals.icu data, including activities, wellness, and calendar events, via natural language.1025 npm1MIT
- AlicenseNot gradedqualityFmaintenanceIntegrates Intervals.icu training data, wellness records, workouts, and calendar events with AI assistants through MCP tools.1MIT
- FlicenseNot gradedqualityDmaintenanceEnables interaction with the Intervals.icu API, providing tools to manage workouts, messages, wellness data, and more through any MCP client.3-
- AlicenseAqualityCmaintenanceMCP server for Intervals.icu that enables AI assistants to manage athletic training data, including activities, calendar events, wellness metrics, and workout libraries.19Apache 2.0