The repo map

The app repository

code/packages/
  backend/     the API, the calc engines, the stores, the enforcement
  frontend/    the web UI
  shared/      types and schemas both sides agree on
  wtcx/        the command-line interface

pm/            the product specifications - one .mdx per surface, ~110 of them
ai/            the assessment rubrics the AI runs against
cli/  mcp/     the CLI and the MCP server

pm/ is the unusual one and the most useful one to read. It is the movement's product specification, in the open: what each surface is for, what rule governs it, what was tried and retired and why. Arguments the movement had with itself are written down there, including the ones it lost.

If you want to understand why something works the way it does, that directory answers it more directly than the code does.

The data repository

problems/         what is wrong, with root causes
laws/             problem_laws/  good_bills/  good_laws/
politicians/      legacy/  new/  both/  worrisom_challengers/  appearances/
qualifications/   the sixteen rubric definitions
scores/           the action catalog, scoring policy, award lists
communities/  geo/  positions/  activism/
provenance/       where every downloaded thing came from
transcriptions/   and the list of what is still missing
schema/           the shapes everything above has to satisfy

Flat files, mostly YAML, readable without our software.

How the two fit together

The app has no hidden copy of the data. It reads the data repository from your disk. That is why a fresh install asks you to clone it first, and why you can inspect and correct anything the app shows you by opening a file.