adr

Run a full Architecture Deep Research pipeline against the current repo and a named decision. Scans the repo for patterns and anti-patterns, drafts a PRD, runs live deep-research with citation audits, and returns a decision handoff the coding agent can implement under.

Run only the discover stage. Scans the local repo for patterns and anti-patterns, drafts a PRD (`pdr.draft.md`), and stops. Use when the user wants to see what ADR infers from the codebase before paying for a full deep-research run.

Set up ADR for first-time use. Audits what's already configured, then walks the user through obtaining and saving the API keys ADR needs. Use when the user asks "set up ADR", "what keys do I need?", "I don't know how to get these", or when /adr:decide reports missing keys.

Install Claude Code + git pre-commit hooks that surface team principles at write time and block commits that violate high-severity rules. Use when the user wants to wire ADR's principles into their everyday workflow so they never see the same review comment twice.

Discover the team's code-review principles from their own repo. Scans for code-review lenses (state-boundaries, llm-call-discipline, etc.), extracts positive patterns + antipatterns + ambiguities with file:line citations, walks the user through an interactive interview to confirm, and writes `.adr/principles.{md,json}`. Use when the user asks to set up team principles, or before running `/adr:review` for the first time.