catence
Catence is a local, read-only MCP server that integrates fitness and health data from Garmin, Intervals.icu, and Strava, enabling AI-assisted analysis and reporting. Key capabilities:
Data Discovery & Status: Check sync state, list datasets, schemas, and coverage via catence_status, describe_data, describe_dataset.
Querying & Aggregation: Fetch time-series data with read_series (with downsampling), run aggregations with aggregate_data, or use custom SQL via query_read_only_data.
Statistical Analysis & Modeling: Perform rolling statistics, correlations, trends, and anomalies with analyze_series; fit regression models with fit_series_model.
Activity & Performance: Find activities by filters with find_activities, retrieve FTP/VO2max history, generate cycling and swim progress reports.
Power Analysis: Explore power-duration bests via power_curve_trend and power_coverage_report.
Swimming: Get lap data with get_swim_laps and session comparisons with swim_progress_report.
Strava Enrichment: On-demand hydration of Strava activities and segments (get_activity_segments, hydrate_strava_activity, etc.).
Search: Find relevant context across activities, plans, and messages with search_context.
All tools are read-only except the explicit hydration operations, ensuring data provenance and quality.
Allows syncing and importing canonical activity and health data from Garmin Connect, including activities, daily health metrics, FTP history, VO2max estimates, and more.
Provides targeted enrichment of activities with Strava data, including hydrating individual activities, recent activities, and segment history, using a read-only Strava API allowlist.
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., "@catence What's my average pace for outdoor runs this year?"
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.
Catence
Local MCP server for primary Garmin data, with added Intervals.icu metrics and Strava enrichment.

... what?
As a context, I'm both an endurance athlete and a data junkie. I already got a coach for all the serious stuff, but I wanted to hook up some—any—LLM to the insane amount of data collected from Garmin, derived on Intervals.icu (I don't pay for TrainingPeaks) and comparable efforts on Strava segments. Naturally, the first answer was a local MCP server for my current tooling.
Related MCP server: fitMCP
What is included
Garmin, Intervals.icu, and Strava ingestion with source-aware normalization.
Read-only MCP tools, plus explicitly named, lock-guarded write tools for Strava hydration and detached syncs.
One shared Catence agent can serve several isolated athlete stores. Every personal-data MCP call names an
athleteId; Catence never silently combines athletes.A password-protected Chainlit Console with an authenticated dashboard, a data-sync button with live progress, and in-app model management.
A generated demo catalog for safe evaluation in Glama, desktop MCP clients, or local development.
Catence helps explore recovery, training load, trends, activity detail, swimming/cycling/running progress, segments, gear, and data quality. It reports the evidence and coverage it can see; it does not diagnose or prescribe training.
Requirements
Node.js 22+
Python 3.12+ and uv for Garmin and Strava provider workers
Provider credentials for the athletes you choose to sync
Quick start
Install Catence and create the first athlete. The default catalog home is ~/.catence; set CATENCE_HOME or pass --home to use a different location.
Every catence-data command requires --athlete <id>; there is no default athlete. Pair it with --home <dir> to select the catalog when not using the default home.
npm install --global catence@latest # or catence@beta
catence-data setup --athlete alex --label "Alex"Store each provider value through stdin so it does not enter shell history. Values are written to the selected athlete's owner-only local secret file.
printf %s 'alex@example.com' | catence-data --athlete alex secret set --provider garmin --field email --value-stdin
printf %s 'your-garmin-password' | catence-data --athlete alex secret set --provider garmin --field password --value-stdin
printf %s 'intervals-api-key' | catence-data --athlete alex secret set --provider intervals --field apiKey --value-stdin
printf %s '12345' | catence-data --athlete alex secret set --provider intervals --field athleteId --value-stdinSync and create retrieval context:
catence-data --athlete alex sync --provider all
catence-data --athlete alex build-retrieval-indexStart the stdio MCP server:
catenceAn agent first calls list_athletes, then includes athleteId with every data tool. This is intentional: a shared agent may access the stores you configured, but no tool implicitly aggregates or crosses between athletes.
Add another athlete
catence-data athlete add --id sam --label "Sam"
printf %s 'sam@example.com' | catence-data --athlete sam secret set --provider garmin --field email --value-stdin
printf %s 'sam-password' | catence-data --athlete sam secret set --provider garmin --field password --value-stdin
catence-data --athlete sam sync --provider garmin
catence-data athlete listGarmin, Intervals, and Strava client credentials are isolated per athlete. Strava OAuth tokens remain in that athlete's own store. The old 0.1 single-store directory cannot be migrated safely: create a fresh home and re-sync each athlete.
Common operations
catence-data --athlete alex status
catence-data --athlete alex sync --provider intervals
catence-data --athlete alex sync --provider garmin --from 2025-07-29
catence-data --athlete alex backfill --provider garmin --from 2020-01-01 --refresh
catence-data --athlete alex retry --run <run-id>
catence-data --athlete alex progress --watch
catence-data --athlete alex auth strava --callback
catence-data --athlete alex disconnect strava
catence-data update --check
catence-data updateDocumentation
Document | Covers |
Install, catalog, secrets, sync, background progress, stdio/HTTP MCP, demo, common operations | |
Local web chat setup, model config, doctor, updates, troubleshooting | |
One-container stack: deploy script, | |
Complete | |
Model providers (OpenAI, Anthropic, OpenAI-compatible/Azure, Opencode Go/Zen), reasoning effort | |
Design rationale, layers, data flow, serving | |
Implementation contract for answering questions, current tools, planned endpoints | |
Release artifacts: npm, PyPI, APM, MCPB |
MCP clients and HTTP
For a packaged installation, point a client at catence:
codex mcp add catence -- catence
claude mcp add --transport stdio catence -- catenceFor a source checkout:
codex mcp add catence -- npm --prefix /absolute/path/to/catence run mcpOptional local Streamable HTTP MCP and dashboard APIs:
catence serve --host 127.0.0.1 --port 8787GET /api/v1/athletes returns IDs and labels only. GET /api/v1/dashboard requires athleteId, for example http://127.0.0.1:8787/api/v1/dashboard?athleteId=alex&days=28. Browser origins must be listed with --allow-origin; the packaged Console instead proxies the dashboard through its authenticated same-origin route. The MCP server has no authentication of its own.
Configuration
config.example.json documents rate limits, Strava budgets, and Console model profiles. Keep per-athlete provider values in catence-data secret set, not in .env or config.json. The full config.json schema —including per-model reasoning effort— is documented in docs/configuration.md and docs/llm-providers.md.
For source development:
npm ci
npm run check
npm test
UV_CACHE_DIR=$PWD/.cache/uv uv run --project console --group dev python -m pytest console/tests -qData and caveats
The current available sources respond to my own used platforms. I used some wrappers:
Strava is only used for segments and gear.
Caveats
There are two caveats to the data fetching, both coming from the fact that this was created to be used by a single person. Accessing the full Garmin API requires applying via a company—which I don't have—and getting Strava data for multiple users is another headache I don't want to get into.
I don't use the official Garmin Connect API. The library uses email/pwd.
Connecting Strava requires the user to be a Strava Premium subscriber and create an API application. The rotating token is stored locally in
<data-dir>/secrets/strava.json.
Available Tools
35 toolsactivity_decouplingDerive aerobic decoupling and GAP for one activityA
Derive aerobic decoupling (Pa:Hr for running using speed/HR, Pw:Hr for cycling using power/HR — drift across steady first/second halves) and grade-adjusted pace (GAP, Minetti 2002 cost model) from stored activity_samples for one activity. Returns null with explanatory caveats when samples are insufficient, walk-heavy, or stop-heavy. Derived values are descriptive only and never overwrite provider-supplied decoupling/GAP. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required), activityId (required), includeGap.
aggregate_dataAggregate cataloged dataA
Declarative aggregation over one cataloged dataset. A timeBucket adds a time_bucket field, which can be used in orderBy. No joins, arbitrary expressions, or file paths. Requires athleteId from list_athletes.
Inputs: dataset (required), metrics (required), athleteId (required), limit, +6 more.
analyze_seriesAnalyze a descriptive seriesC
Run deterministic rolling statistics, baselines, correlations, seasonal comparisons, or trends on a cataloged series. Requires athleteId from list_athletes.
Inputs: dataset (required), metrics (required), analysis (required), athleteId (required), +9 more.
catence_statusCatence data statusA
Read sync state, data coverage, entity counts, stream availability, unresolved errors, and retrieval-index freshness. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required).
catence_sync_progressCatence sync progressA
Read live progress heartbeats for active sync runs, plus the most recent completed or interrupted runs. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required).
cycling_progress_reportBuild a cycling progress reportA
Combine source-aware FTP and VO₂max histories with monthly canonical volume/load and labelled power-curve trends. Read-only and descriptive. Requires athleteId from list_athletes.
Inputs: athleteId (required), endDate, startDate.
describe_dataDescribe available Catence datasetsA
List cataloged datasets, fields, units, permitted filters/groupings, providers, and time coverage. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required).
describe_datasetDescribe one Catence datasetA
Read a compact schema, permitted filters/groupings, provenance fields, and coverage for one cataloged dataset. Read-only. Requires athleteId from list_athletes.
Inputs: dataset (required), athleteId (required).
find_activitiesFind activities and likely racesA
Find canonical activities by sport, distance, name, and date. Results are paginated and transparently flag likely-race signals without claiming provider-confirmed race metadata. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required), sort, limit, sport, +6 more.
fit_series_modelFit a descriptive series modelB
Fit a bounded OLS, Theil–Sen, quadratic, or cubic descriptive model. Not a sport-performance model. Requires athleteId from list_athletes.
Inputs: model (required), dataset (required), metrics (required), athleteId (required), +9 more.
get_activity_segmentsHydrate and read an activity’s Strava segmentsA
For a selected activity’s segments, climbs, KOM/PRs, or per-segment analysis: automatically hydrate the matching Strava activity before returning its persisted segment efforts. Use this before saying segment data is unavailable. Requires athleteId from list_athletes.
Inputs: athleteId (required), activityId (required), limit, refresh.
get_ftp_historyGet dated FTP historyB
Return normalized cycling FTP settings and activity-summary observations with source-aware preferred daily values. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required), sport, endDate, startDate, +1 more.
get_swim_lapsRead explicit swim lengths and grouped setsA
Return source-aware swim lengths when a provider actually supplied them, plus Garmin detected and Intervals.icu auto-detected sets. A missing length list is reported as unavailable; laps are never reconstructed from samples. Requires athleteId from list_athletes.
Inputs: athleteId (required), activityId (required), provider.
get_vo2max_historyGet sport-specific VO₂max historyA
Return normalized Garmin VO₂max observations for exactly one sport. For running, pass sport: running (or run): Garmin supplies that series with raw sport generic, which is preserved in each row. Cycling remains separate. Omit sport only to inspect available source labels. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required), sport, endDate, startDate.
hydrate_recent_strava_activitiesHydrate several recent Strava activitiesA
Write-only targeted batch enrichment. Select an explicit list or a bounded date/sport window; Catence hydrates one activity at a time and reports every outcome. Requires athleteId from list_athletes.
Inputs: athleteId (required), limit, sports, endDate, +3 more.
hydrate_strava_activityHydrate one activity from StravaA
Write-only targeted enrichment. Safely matches exactly one Catence activity/source to Strava, then archives and persists its activity detail, gear assignment, and segment efforts. For a segment/climb request, prefer get_activity_segments, which invokes this prerequisite automatically. Requires athleteId from list_athletes.
Inputs: athleteId (required), activityId (required), refresh.
hydrate_strava_segment_historyHydrate one Strava segment historyA
Write-only targeted enrichment. Fetches a persisted Strava segment and the authenticated athlete’s paged historic efforts; interrupted or throttled work is resumable. Requires athleteId from list_athletes.
Inputs: athleteId (required), segmentId (required), refresh.
latest_cycling_activitiesList latest cycling activitiesA
List Garmin cycling source records without duplicate Intervals summaries, optionally flagging multisport parent records. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required), limit, endDate, startDate, +1 more.
list_athletesList configured athletesA
List athlete IDs and labels available to this shared Catence agent. No personal metrics are returned.
Inputs: no named parameters.
power_coverage_reportInspect FIT-derived power coverageA
Report powered activities and the complete duration-best inventory for an explicit sport or sport family. Uses Garmin FIT-derived power, not sparse activity summaries. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required), sport, endDate, startDate, +2 more.
power_curve_trendRead labelled power-curve trendsA
Return monthly bests for selected power durations from an explicit sport or sport family, including the supporting activity and source type. Read-only. Requires athleteId from list_athletes.
Inputs: athleteId (required), sport, endDate, durations, +3 more.
query_read_only_dataQuery cataloged read-only dataA
Advanced fallback after describe_data/describe_dataset: one parameterized SELECT or WITH … SELECT over cataloged views only. Do not query information_schema, DuckDB system tables, or uncataloged physical relations. Results use deterministic cursor pagination and always report whether they are incomplete. Filesystem access, extensions, DDL, and mutation are rejected. Requires athleteId from list_athletes.
Inputs: sql (required), athleteId (required), cursor, values, +1 more.
readiness_baselineRead a running/race readiness baselineA
Combine the full performance-indicator set — running lactate threshold (pace/power/HR), VO₂max, race predictions, fitness trend, endurance score, and FIT-derived power bests/coverage — as trends in one call. Use this before any single-activity comparison for race-readiness or fitness-target questions. Requires athleteId from list_athletes.
Inputs: athleteId (required), sport, endDate, startDate.
read_seriesRead a bounded time seriesA
Read cataloged numeric series with deterministic cursor pagination and automatic stream downsampling. metrics must be numeric catalog columns; place identifiers and other strings in filters. Call describe_dataset first if the fields are uncertain. Requires athleteId from list_athletes.
Inputs: dataset (required), metrics (required), athleteId (required), cursor, +6 more.
resolve_event_courseResolve a race event course and its synced geometryA
Given a Garmin/Intervals eventId, return the courseId it references and whether that course geometry has been synced. When geometry is absent, the result carries an explicit caveat so a prior course profile is never reused. Optionally pass pastActivityId to diff the event course against that activity's course. Requires athleteId from list_athletes.
Inputs: eventId (required), athleteId (required), pastActivityId.
review_activity_deep_diveReview one activity in depthA
Return canonical activity identity, source summaries, and structured intervals. Call read_series separately only when a specific sampled metric is required. Requires athleteId from list_athletes.
Inputs: athleteId (required), activityId (required).
review_daily_recovery_loadReview daily recovery and training loadA
Return source-cited daily health, training, and nutrition facts for a recovery/load review. It does not prescribe a training plan. Requires athleteId from list_athletes.
Inputs: athleteId (required), date.
review_weekly_trainingReview seven days of trainingA
Return source-cited health, training, and nutrition facts for the seven-day period ending on an optional date. It does not prescribe a training plan. Requires athleteId from list_athletes.
Inputs: athleteId (required), endDate.
search_contextSearch generated contextA
Search compact generated activity, plan, nutrition, and message context. Results identify authoritative follow-up tools rather than making numerical claims. Requires athleteId from list_athletes.
Inputs: query (required), athleteId (required), limit, filters.
start_detached_syncStart a detached data syncA
Write-only sync trigger, named like the other explicit lock-guarded write tools. Spawns one detached catence-data sync process for this athlete and returns immediately with the run handle and log file; it refuses while another sync run is active. Track the run with catence_sync_progress; this tool never waits for completion. Requires athleteId from list_athletes.
Inputs: athleteId (required), from, refresh, provider.
swim_progress_reportBuild a source-aware swim progress reportA
Compare Garmin swimming summaries and data completeness over a date range, optionally within one pool length. It does not derive pace trends from moving time divided by distance. Requires athleteId from list_athletes.
Inputs: athleteId (required), endDate, startDate, poolLengthM.
wellness_anomaliesFind statistical wellness anomaliesB
Find daily recovery and wellness outliers by z-score and group dates with anomalies across multiple signals. This does not diagnose or prescribe. Requires athleteId from list_athletes.
Inputs: athleteId (required), endDate, metrics, startDate, +1 more.
wellness_baselinesRead personal wellness baselinesA
Return trailing means, standard-deviation bands, latest values, and latest z-scores for common recovery and training metrics. Missing values remain missing. Requires athleteId from list_athletes.
Inputs: athleteId (required), endDate, metrics, windowDays.
wellness_correlateCorrelate recovery and training metricsA
Calculate a compact daily Pearson or Spearman correlation between curated wellness and training metrics, with an optional -7 through +7 day lag scan. Descriptive only. Requires athleteId from list_athletes.
Inputs: metricA (required), metricB (required), athleteId (required), method, +4 more.
wellness_coverageInspect wellness data coverageA
Report present and missing dates for curated wellness/training metrics, plus unresolved extraction errors. Missing data is never interpreted as rest or a health outcome. Requires athleteId from list_athletes.
Inputs: athleteId (required), endDate, metrics, startDate.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 35 tools
Most tools map cleanly to distinct resource/action pairs (e.g., describe_data vs describe_dataset, read_series vs aggregate_data). A few pairs like readiness_baseline/wellness_baselines and catence_status/catence_sync_progress could be confused, but their descriptions generally clarify the boundary.
The vast majority of tools follow a consistent snake_case verb_noun pattern like review_*, get_*, describe_*, and hydrate_*. Minor deviations such as activity_decoupling, wellness_correlate, and catence_status keep it from a perfect 5, but the overall naming is predictable.
At 35 tools, the surface is large and exceeds the 'too many' threshold for most agent-facing servers. Even for a broad athlete analytics domain, the selection burden is significant, and several related reporting tools could plausibly be consolidated.
The set covers the apparent athlete-analytics domain well: activity retrieval, series querying, statistical analysis, wellness metrics, cycling/swimming/running reports, sync control, and Strava enrichment. Minor gaps exist, such as no direct plan or nutrition detail tool despite search_context indexing those concepts, but core workflows generally resolve without dead ends.
Maintenance
Related MCP Connectors
Strava MCP tools for AI: athletes, activities, segments, clubs, routes. Powered by HAPI MCP server.
Adaptive running coach MCP server — training data, plans, and recovery for AI assistants.
Remote MCP server for training, nutrition, wellness, and performance data with OAuth 2.0.
MCP server for Withings health data — sleep, activity, heart, and body metrics.
Related MCP Servers
- AlicenseAqualityDmaintenanceMCP server that exposes Garmin Connect health and activity data (steps, sleep, stress, activities, etc.) via tools for querying, analysis, and visualization.17Apache 2.0
- FlicenseNot gradedqualityDmaintenanceA multi-platform fitness MCP server that syncs data from Garmin, Strava, Google Fit, and Suunto into a local DuckDB database and provides analytics tools via MCP.1-
- AlicenseBqualityBmaintenancePrivacy-first MCP server for Strava activities, streams, routes and training data.29235 npm3MIT
- AlicenseAqualityDmaintenanceMCP server for reading and querying Garmin Connect data, including activities, strength history, recovery, trends, and optionally creating workouts.12MIT