The Unit coverage badge in the README is code coverage: the share of this project's own source lines — and of the branches through them — that the automated test suite actually executes when it runs. CI measures it with coverage.py on every push and pull request, and fails the build if it drops below 90%.
Read it as a floor, not a grade. It says how much of the code is exercised by tests at all; it does not say the tested parts are correct, since a line can be run by a test that never checks the result. What it does tell you is where nothing is watching: uncovered code has never been run by a test, so no test would notice if a change broke it. That matters here because this tool creates and edits GitHub repositories on your behalf, and most of what it does is impractical to verify by hand on every change.
The badge shows today's number. The full history is a CSV with one row per push to main, if you would rather see the trend than a snapshot. And unit coverage is only one of five separate measurements this project tracks, which are not interchangeable — CONTRIBUTING.md explains what each one covers and how to reproduce it locally.