Ek Rasta

Some government documents are impossible to obtain because the requirements are circular. In Madhya Pradesh you need a proof of residence to get a ration card, and the ration card is itself listed as an accepted proof of residence. There is no entry point. Citizens who hit this assume they made a mistake, return to the office repeatedly, and eventually give up. The departments cannot see it either, because each office only sees its own form. The trap only appears when requirements from different departments are placed side by side, which nobody currently does.

Ek Rasta maps government paperwork as a directed graph instead of a checklist. You name an outcome, such as a new ration card. Claude Fable 5.1 generates every document that outcome requires, recursively, with a stated basis for each requirement. The application then runs Tarjan's strongly connected components algorithm over that graph in our own code to find every circular dependency. The loops are computed, not asserted by the model, which means the finding is verifiable rather than a claim.

We tested six Madhya Pradesh processes. All six contained at least one circular dependency. The ration card graph has 14 documents and 27 requirements, with 3 circular pairs sitting inside one deadlocked group of 9 documents. The tool is built for the department, not the citizen. A citizen already knows they are stuck. The department does not, and it is the only party that can remove the requirement causing the deadlock.

One known limitation, which we flag in the interface: some nodes are requirement categories rather than specific documents, such as proof of residence, where any one of several documents is accepted. Loops through those may be satisfiable another way, and we label them accordingly. Loops between two specific named documents are marked without that caveat.

Claude Fable 5.1 generates every document that outcome requires, recursively, with a stated basis for each requirement.