Model Context Protocol
Model Context Protocol
Model Context Protocol
Things you can ask once connected
MCP is a Plus, Pro & Elite feature
Setup
Connect via the Connectors panel — no config files, no restarts.
Open Customize → Connectors in Claude.ai.
Click the + button, then Add custom connector.
Paste the MCP server URL.
https://leaderboard.y11k.com/api/mcpSign in to Leaderboard in the OAuth pop-up and approve the read scopes.
Try it.
Ask Claude: “What’s my current rank on my leaderboards?”
Available on Free, Pro, Max, Team, and Enterprise plans. Free accounts are limited to one custom connector; Team/Enterprise members may need an admin to allow custom connectors first.
Capabilities
Twenty read-only tools — your activities, leaderboards, and events. Your assistant can only call what you’ve approved.
read:activitiesget_my_activitiesRecent Strava activities — name, type, distance, duration, elevation.
search_my_activitiesFilter by date, type, distance, moving time, or name. Sortable + paginated.
get_activity_detailFull detail of a single activity, including pace and map polyline.
get_my_statsWeekly rolling stats with deltas, current streak, and lifetime totals.
get_my_personal_recordsLongest distance, biggest elevation, longest moving time per activity type.
get_my_personal_bestsYour current best time per named effort — from 400m to marathon, as shown on the trophies page.
get_my_activity_breakdownAggregate by week, month, type, or day-of-week over any date range.
get_my_training_loadPer-week load, ramp %, monotony, and phase (build / steady / overload / taper).
get_my_month_recapCached month-in-review snapshot for any month (default: previous month).
get_my_year_recapCached year-in-review snapshot (default: current year, YTD).
read:leaderboardslist_my_leaderboardsAll leaderboards you belong to, with your current rank in each.
get_leaderboard_rankingsRanked members of a leaderboard, by metric and time window.
get_leaderboard_activitiesRecent activities by members of a leaderboard, cursor-paginated.
get_leaderboard_timeseriesPer-member daily cumulative series plus a 30-day forecast band.
get_leaderboard_eventsRecent rank changes, joins, leaves, and milestones.
get_rank_historyYour daily rank trajectory on a leaderboard over the last N days.
get_rank_gapGap to a target rank, with the daily pace needed to close it and an ETA.
get_member_profileResolve a member’s display name and avatar from their user ID.
read:eventslist_my_eventsRaces, group runs and virtual challenges you’re taking part in, with registration status (approved / pending / waitlisted / declined).
get_event_resultsResults of one event — rank, time, percentile, personal-best flag. Respects each participant’s privacy setting.
Ready-made prompts
Thirteen built-in prompts ship with the server — copy one into any connected assistant to get started.
Security
read:activitiesRead your Strava activities, weekly stats, and streak.read:leaderboardsRead leaderboard rankings, members, and recent events you have access to.read:eventsRead races, group runs and virtual challenges you’re taking part in, and their results.offline_accessRefresh access in the background so you don’t have to re-authorize daily.Access tokens last 1 hour; refresh tokens last 30 days. Revoking a connection invalidates both immediately.
Beyond the tools above, the server also exposes background resources — user://me, activity://{id}, leaderboard://{id} — so an assistant can pull your profile, a single activity, or a leaderboard without a tool call.
FAQ
Tool responses are returned to the AI assistant that called them, the same way any response in your chat would be. We don’t share data with anyone else, and the assistant has no background access — it can only call tools when you’re actively chatting.
No. Every tool exposed over MCP is read-only. There’s no way for an assistant to post activities, edit leaderboards, invite members, or modify any setting.
The assistant loses access immediately. Both its access and refresh tokens are invalidated server-side, so any cached state in the client stops working on the next request.
Each AI assistant runs its own OAuth client. You approve which scopes each one can see, and you can revoke them independently from Settings → MCP Connections.