adr-review

name	adr-review
description	Review an ADR that Zoha has written in docs/adr/. Use when Zoha says an ADR is ready for review, or asks for feedback on a decision record. Never use this to write or rewrite an ADR.

Reviewing a dinotable ADR

ADRs are Zoha's writing — that constraint is the point of the exercise. You review; you never draft, rewrite, or "clean up" the ADR text. The only file you may edit is the index table in docs/adr/README.md.

Review checklist

Read the ADR (and the template docs/adr/0000-template.md), then assess:

1. Context — would Zoha-in-six-months understand why this was a question? Are the real constraints named (team-of-one, learning goals, Dhaka-scale numbers from design doc §5)?
2. Options — are at least two real options presented? Is each option's strongest case made fairly, or is one a strawman? Is "defer / do nothing" considered where legitimate?
3. Decision — does it name the one or two factors that actually tipped it, or does it just restate all the pros?
4. Consequences — are the costs of the chosen option stated honestly? Is there a concrete revisit-trigger ("if X happens, reopen this")?
5. Consistency — does it contradict the design doc or a prior ADR? If it changes the plan, flag that the design doc needs a matching update (by Zoha or by you — design doc edits are allowed).

How to deliver the review

• Start with what the ADR gets right — specifically, not generically.
• Raise issues as questions where possible ("what would make you reopen this?") rather than corrections. Socratic first, direct second.
• Distinguish blocking problems (unfair option framing, missing consequences) from polish (wording, formatting). Don't nitpick prose.
• End with a clear verdict: accept as-is / accept with noted gaps / worth another pass — and if accepted, update the status in the ADR's own header is Zoha's edit, but update the index row in docs/adr/README.md yourself.
• If you disagree with the decision itself but the reasoning is sound and the trade-off is understood: say so, then accept it. A defended decision Zoha owns beats a correct decision Zoha was handed.