MuscleScript: From prompt to pump.

Choose from top models under health category. Complete transparency which most apps don't provide what they use under the hood. They call it our AI but mostly it's a wrapper.

Comment

Inspiration

  • I generated my workout plan from claude code. Now everytime I start my workout I need to navigate to artifacts section and check my workout table, find workout for the day. This is painful. Even if I download it as pdf file still with new files in my system will move my workout file below. Even if I pin my workout file in file manager I will still have to find workout for the day in the table.

  • I want to make a workout planner app just like we have cursor for coding. I want to ability to see today's workout, track my work out and count calories. There are many other apps in the market but they are not transparent about what AI model they use. I want to use open router's service so that users can choose their own AI Model. Users will have freedom to put their own api key or they can purchase the subscription plan.

What it does

  • It helps techies help plan their workout using AI.
  • It can help anyone who want to try different workout routines or someone who easily gets bored of same kind of workout or self learners.
  • App lets users track their workout generated by top AI models. Most app keeps preset of data then generate workout plan. This app does it differently. You can generate any kind of routine you can think of based upon your target and available assets.

How we built it

  • Will use react native with open router api to let users choose their own AI Model.
  • It also uses revenuecat for ease of subscription handling
  • I build it as solo dev did design, frontend and backend all by myself and with help of AI coding using claude code.

Challenges we ran into

  • Results might be inconsistent based upon AI model's performance that is why I am only using AI models which are listed under health category in open router.
  • Making AI system prompt is such a way it never generates out of context response. I validate response from AI and restrict it to output response in certain format using which I can help users track their workout.

Accomplishments that we're proud of

  • I have been able to solve my own use case and I believe it can help others also in their workout journey.

What we learned

  • It takes a lot to build and ship a polished product live. AI technology helps but still manual intervention is needed. Without professional experience of 7+ years into app development this was not possible for me.

What's next for MuscleScript → your body’s workout script.

  • Diet planning using AI models based upon user preferences providing complete transparency unlike most app in market which are not transparent about which ai model they use and which are limited by their preset of database.
  • Image/gif or video generation perfect workout providing users clean muscle groups. All I will keep a common database of all exercises videos and feed it into ai response using MCP server at backend. It might miss some of the exercises. For those we will not render video we will show button to view workout form over youtube(Which I usually do when I am working out in GYM)
Share this project:

Updates