# MuscleCloud Product and Connection Reference

MuscleCloud is a gym tracker for people who want to keep a useful training record and optionally connect their own AI assistant. It supports logging routines, exercises, sets, reps, weights, completed sessions, training dates, and streaks. Previous values follow the same exercise across routines.

## What people can do

- Log workouts without using AI.
- Ask their connected assistant to compare lifts, review consistency, or explore exercise history.
- Work with their assistant on routines suited to their goals, equipment, and schedule.
- Let their assistant search the exercise catalogue and create or update routines.
- Identify AI-created or updated routines by the connection's nickname.

Analysis and suggestions come from the connected assistant, not a built-in MuscleCloud AI coach. No AI subscription is included. MuscleCloud is not a substitute for a qualified trainer or clinician and cannot assess physical exercise form.

## Public URLs

- Home: https://www.musclecloud.app/
- Connection guide: https://www.musclecloud.app/ai-access.html
- MCP endpoint: https://api.musclecloud.app/mcp
- Read-only context endpoint: https://api.musclecloud.app/ai/context
- API base: https://api.musclecloud.app

## User connection flow

Sign in, open Share with AI, enter an AI nickname, and choose Copy Connection Instructions. This creates a token and copies the token, MCP URL, and context URL. Manual mode provides individual controls. Previously issued tokens can be named and revoked.

The assistant must support remote MCP with bearer-token authentication or authenticated HTTP requests. Compatibility depends on the assistant and its settings. Nicknames are labels, not supported-provider guarantees.

## MCP tools

- get_musclecloud_context: read workout history, routines, training dates, exercise history, and consent guidance.
- search_exercises: search the catalogue by name, muscle, or equipment.
- get_exercise_details: retrieve details for a catalogue exercise.
- create_routine: create a routine for the authenticated user.
- update_routine: update an existing routine for the authenticated user.

The context endpoint is a read operation returning JSON. It does not itself make changes. The connection also permits routine creation and updates through the MCP tools and authenticated routine API. It is not read-only access overall. Completed workout editing is handled in the web app, not by these MCP tools.

## Consent and security

Before creating a routine, explicitly confirm that the user wants it created.
Before modifying a routine, obtain explicit permission for the proposed changes.
Explain the routine name, exercises, sets, reps, and material changes before calling a write tool.
These instructions guide the assistant; the app does not enforce a separate approval dialog.

Account endpoints require Authorization: Bearer <token>. Tokens are secrets and grant access to the issuing user's training data and routine tools. Do not put tokens in public pages or URLs. Use separate named connections for separate assistants. Revoking a token stops future access but cannot remove data already shared with an AI provider. The provider's data policies apply.

Exercise data: RepDB, https://repdb.co/