On this page
codeps 🔗
codeps tracks broad dependency-health trends for JVM projects. It reads compiler output, identifies cycles and coupling signals, keeps compact history, and generates a static dashboard for each configured project.
Configure the repository once, then run one command in CI:
brew install sake92/tap/codeps
codeps status
The examples below use codeps as the installed CLI command.
The command writes:
.codeps/<project>.ndjson— compact package/file health history intended for Git..codeps/out/<project>/report.json— latest detailed report..codeps/out/<project>/index.html— interactive D3/Pico dashboard.
Start with codeps findings --project <id> --limit 10 --format json, then use codeps inspect-cycle or codeps inspect-node for selected subjects, preserving the project and scope. Pass --scope files for file detail. The report is a trend signal, not a substitute for architectural judgment: its metrics are intended to be good enough to reveal a direction of travel.
See the CLI reference for the repository configuration and GitHub Pages workflow, Tutorials for guided setup, and How Tos for SemanticDB and jdeps input. You can also use the codeps example projects as ready-made repositories to try, and the codeps skill with an AI coding agent. See Use codeps with an AI agent for installation and example prompts.