Loading
About
What this project is, how it is built, and where it deliberately stops.
Manomithra arranges yoga practices into routines for a single person. You say what you want from your practice, how much time you have, and what your body can comfortably do; it chooses from a documented library and puts the pieces in an order you can follow.
The practice library it draws from is made of Manomithra's own records: asana, mudra and pranayama, each one written, sourced and reviewed before it can appear in a routine.
It is straightforward to ask a language model to invent a yoga routine, and the result will usually read well. The problem is that a well-written instruction that is subtly wrong is worse than no instruction at all, because a reader has no way to tell.
So the content is separated from the arrangement. Content is authored, attributed and reviewed by a person. Arrangement is computed. The two never mix, and no part of the routine is written at request time.
The system is six separate pieces with clear boundaries: the practice library, the safety rules, the interpretation step that reads your answers, the routine planner, the session you practise from, and the feedback loop that informs the next routine. Each has its own interface, and the boundaries are enforced in the data model rather than left to convention.
That structure is not incidental. It is what makes the claims checkable, and it is what will let the pieces be tested one at a time.
Manomithra does not diagnose, does not give medical advice, and does not interpret symptoms. It does not know your history, and it will not ask you to tell it things it should not be told.
It does not invent practices. It does not claim any outcome. It does not currently contain the full practice library, and the pages that depend on that content say so rather than filling the gap.
Where a question needs a person — a medical practitioner, or a teacher who can see you practise — the product says so and stops.
Next
The practice library, both salutation sequences and the safety approach are all open without an account. That is the fastest way to judge whether this is built the way it claims to be.