Look back.
Review weights, reps, completed sessions, and consistency.
“Which lifts have improved over the last few weeks?”
Bring your own AI
Your training joins
the conversation.
Give your assistant something useful to work with: your actual workouts. Ask for a closer look at your progress, a fresh routine, or a small adjustment to the one you love.
A simple connection.
Use an assistant with MCP support or authenticated HTTP access. Connection options vary by assistant and account.
Name your AI.
In MuscleCloud, open Share with AI and give the connection a nickname.
Copy the instructions.
Choose Copy Connection Instructions. This creates a token and copies the details your assistant needs.
Start a conversation.
Give the instructions to your assistant using its connection setup. Ask it to review your training or work on a plan with you.
Not one fixed way to train
A few places to start.
Bring your own questions, or ask your AI what might be worth exploring.
Look ahead.
Explore the exercise catalogue and build a routine around your constraints.
“I have 30 minutes and a pair of dumbbells. What could we plan?”
Adjust together.
Discuss changes, then ask your assistant to update a saved routine.
“Suggest a shorter version of my routine. Ask me before changing it.”
Know what you share
Access, with clarity.
What can my AI access?
A connected assistant can read your routines and workout history, including exercises, sets, reps, weights, training dates, and muscle groups. It can explore the exercise catalogue and create or update your routines. AI-created or updated routines show the connection’s nickname.
Will it ask before making changes?
The context and tool instructions tell assistants to get your explicit permission before creating or changing routines. This is guidance for the assistant, not an approval dialog enforced by MuscleCloud. Only connect assistants you trust.
Can I disconnect an assistant?
Yes. Revoke its named token in Share with AI. That stops future requests using the token, but does not remove information the assistant has already received.
How should I handle connection instructions?
They contain a secret token that grants access to your account’s training data and routine tools. Keep it private, use a separate connection for each assistant, and revoke it if it is exposed. Your AI provider’s data policies apply to anything you share with it.
Does every AI app support this?
No. The assistant must support remote MCP with bearer-token authentication or authenticated HTTP requests. A nickname such as ChatGPT or Claude is a label, not a guarantee of compatibility. MuscleCloud does not include an AI subscription or provide its own AI coach.
For assistants and developers
The connection details.
MuscleCloud supports Model Context Protocol (MCP) and an HTTP context API. Both require a user-issued token in the Authorization: Bearer <token> header. Public pages never expose a user’s training data.
- MCP endpoint
https://api.musclecloud.app/mcp- Context endpoint
https://api.musclecloud.app/ai/context
The context endpoint returns training data and routine-change guidance as JSON. It is a read operation. To create or update routines, use the MCP tools or the authenticated routine API.
get_musclecloud_context: read training history, routines, and consent guidance.search_exercises: search by name, muscle, or equipment.get_exercise_details: inspect an exercise from the catalogue.create_routine: create a routine after asking the user.update_routine: change a routine after getting explicit permission.